[InstructionCost] Don't conflate Invalid costs with Unknown costs.

We previously made a change to getUserCost to return a Invalid cost
when one of the TTI costs returned '-1' (meaning 'unknown' or
'infinitely expensive'). It makes no sense to say that:

  shufflevector <2 x i8> %x, <2 x i8> %y, <4 x i32> <i32 0, i32 1, i32 2, i32 3>

has an invalid cost. Perhaps the cost is not known, but the IR is valid
and can be code-generated. Invalid should only be used for IR that
cannot possibly be code-generated and where a cost is nonsensical.

With more passes now asserting that the cost must be valid, it is possible
that those assertions will fail for perfectly valid IR. An incomplete
cost-model probably shouldn't be a reason for the compiler to break.

It's better to consider these costs as 'very expensive' and ignore them
for other reasons. At some point, we should consider replacing -1 with
some other mechanism.

Reviewed By: paulwalker-arm, dmgreen

Differential Revision: https://reviews.llvm.org/D99502

diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index 23f26fe..87b36ac 100644 (file)
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -237,8 +237,6 @@ public:
       Cost = getUserCost(I, kind);
       break;
     }
-    if (Cost == -1)
-      Cost.setInvalid();
     return Cost;
   }
 

diff --git a/llvm/lib/Analysis/CostModel.cpp b/llvm/lib/Analysis/CostModel.cpp
index 19c307b..83b7d5c 100644 (file)
--- a/llvm/lib/Analysis/CostModel.cpp
+++ b/llvm/lib/Analysis/CostModel.cpp
@@ -107,7 +107,7 @@ void CostModelAnalysis::print(raw_ostream &OS, const Module*) const {
       if (auto CostVal = Cost.getValue())
         OS << "Cost Model: Found an estimated cost of " << *CostVal;
       else
-        OS << "Cost Model: Unknown cost";
+        OS << "Cost Model: Invalid cost";
 
       OS << " for instruction: " << Inst << "\n";
     }

diff --git a/llvm/test/Analysis/CostModel/AMDGPU/br.ll b/llvm/test/Analysis/CostModel/AMDGPU/br.ll
index 494f8d2..9c85cca 100644 (file)
--- a/llvm/test/Analysis/CostModel/AMDGPU/br.ll
+++ b/llvm/test/Analysis/CostModel/AMDGPU/br.ll
@@ -20,7 +20,7 @@ bb2:
 }
 
 ; CHECK: 'test_switch_cost'
-; CHECK: Unknown cost for instruction:   switch
+; CHECK: estimated cost of -1 for instruction:   switch
 define amdgpu_kernel void @test_switch_cost(i32 %a) #0 {
 entry:
   switch i32 %a, label %default [

diff --git a/llvm/test/Analysis/CostModel/AMDGPU/shufflevector.ll b/llvm/test/Analysis/CostModel/AMDGPU/shufflevector.ll
index ff3abca..a276051 100644 (file)
--- a/llvm/test/Analysis/CostModel/AMDGPU/shufflevector.ll
+++ b/llvm/test/Analysis/CostModel/AMDGPU/shufflevector.ll
@@ -53,7 +53,7 @@ define amdgpu_kernel void @shufflevector_02_v2i16(<2 x i16> addrspace(1)* %out,
 }
 
 ; GCN-LABEL: 'shufflevector_xxx'
-; TPT: Unknown cost for {{.*}} shufflevector <2 x i8> %vec, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; TPT: estimated cost of -1 for {{.*}} shufflevector <2 x i8> %vec, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
 ; CS: estimated cost of 1 for {{.*}} shufflevector <2 x i8> %vec, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
 ; Should not assert
 define amdgpu_kernel void @shufflevector_xxx(<4 x i8> addrspace(1)* %out, <2 x i8> addrspace(1)* %vaddr) {

diff --git a/llvm/test/Analysis/CostModel/ARM/shuffle.ll b/llvm/test/Analysis/CostModel/ARM/shuffle.ll
index 65f7ea0..0cefb4a 100644 (file)
--- a/llvm/test/Analysis/CostModel/ARM/shuffle.ll
+++ b/llvm/test/Analysis/CostModel/ARM/shuffle.ll
@@ -171,39 +171,39 @@ define void @reverse() {
 
 define void @concat() {
 ; CHECK-MVE-LABEL: 'concat'
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v4i8 = shufflevector <2 x i8> undef, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v8i8 = shufflevector <4 x i8> undef, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v16i8 = shufflevector <8 x i8> undef, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v4i16 = shufflevector <2 x i16> undef, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v8i16 = shufflevector <4 x i16> undef, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v16i16 = shufflevector <8 x i16> undef, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v4i32 = shufflevector <2 x i32> undef, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v8i32 = shufflevector <4 x i32> undef, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v4i64 = shufflevector <2 x i64> undef, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v4f16 = shufflevector <2 x half> undef, <2 x half> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v8f16 = shufflevector <4 x half> undef, <4 x half> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v16f16 = shufflevector <8 x half> undef, <8 x half> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v4f32 = shufflevector <2 x float> undef, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v8f32 = shufflevector <4 x float> undef, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-MVE-NEXT:  Cost Model: Unknown cost for instruction: %v4f64 = shufflevector <2 x double> undef, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4i8 = shufflevector <2 x i8> undef, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8i8 = shufflevector <4 x i8> undef, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v16i8 = shufflevector <8 x i8> undef, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4i16 = shufflevector <2 x i16> undef, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8i16 = shufflevector <4 x i16> undef, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v16i16 = shufflevector <8 x i16> undef, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4i32 = shufflevector <2 x i32> undef, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8i32 = shufflevector <4 x i32> undef, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4i64 = shufflevector <2 x i64> undef, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4f16 = shufflevector <2 x half> undef, <2 x half> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8f16 = shufflevector <4 x half> undef, <4 x half> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v16f16 = shufflevector <8 x half> undef, <8 x half> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4f32 = shufflevector <2 x float> undef, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8f32 = shufflevector <4 x float> undef, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4f64 = shufflevector <2 x double> undef, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 ; CHECK-MVE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; CHECK-NEON-LABEL: 'concat'
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v4i8 = shufflevector <2 x i8> undef, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v8i8 = shufflevector <4 x i8> undef, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v16i8 = shufflevector <8 x i8> undef, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v4i16 = shufflevector <2 x i16> undef, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v8i16 = shufflevector <4 x i16> undef, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v16i16 = shufflevector <8 x i16> undef, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v4i32 = shufflevector <2 x i32> undef, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v8i32 = shufflevector <4 x i32> undef, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v4i64 = shufflevector <2 x i64> undef, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v4f16 = shufflevector <2 x half> undef, <2 x half> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v8f16 = shufflevector <4 x half> undef, <4 x half> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v16f16 = shufflevector <8 x half> undef, <8 x half> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v4f32 = shufflevector <2 x float> undef, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v8f32 = shufflevector <4 x float> undef, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEON-NEXT:  Cost Model: Unknown cost for instruction: %v4f64 = shufflevector <2 x double> undef, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4i8 = shufflevector <2 x i8> undef, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8i8 = shufflevector <4 x i8> undef, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v16i8 = shufflevector <8 x i8> undef, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4i16 = shufflevector <2 x i16> undef, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8i16 = shufflevector <4 x i16> undef, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v16i16 = shufflevector <8 x i16> undef, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4i32 = shufflevector <2 x i32> undef, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8i32 = shufflevector <4 x i32> undef, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4i64 = shufflevector <2 x i64> undef, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4f16 = shufflevector <2 x half> undef, <2 x half> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8f16 = shufflevector <4 x half> undef, <4 x half> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v16f16 = shufflevector <8 x half> undef, <8 x half> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4f32 = shufflevector <2 x float> undef, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v8f32 = shufflevector <4 x float> undef, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %v4f64 = shufflevector <2 x double> undef, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 ; CHECK-NEON-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %v4i8 = shufflevector <2 x i8> undef, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>

diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-extract_subvector.ll b/llvm/test/Analysis/CostModel/X86/shuffle-extract_subvector.ll
index 29eb5a4..ffc7aba 100644 (file)
--- a/llvm/test/Analysis/CostModel/X86/shuffle-extract_subvector.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-extract_subvector.ll
@@ -27,7 +27,7 @@ define void @test_vXf64(<4 x double> %src256, <8 x double> %src512) {
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_2345 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_4567 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-; SSE-NEXT:  Cost Model: Unknown cost for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 undef>
+; SSE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 undef>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf64'
@@ -40,7 +40,7 @@ define void @test_vXf64(<4 x double> %src256, <8 x double> %src512) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V512_2345 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_4567 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 undef>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf64'
@@ -53,7 +53,7 @@ define void @test_vXf64(<4 x double> %src256, <8 x double> %src512) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_2345 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_4567 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 undef>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V256_01 = shufflevector <4 x double> %src256, <4 x double> undef, <2 x i32> <i32 0, i32 1>

diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector.ll b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector.ll
index b50f7f9..f242692 100644 (file)
--- a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector.ll
@@ -18,9 +18,9 @@
 
 define void @test_vXf64(<2 x double> %src128, <4 x double> %src256, <8 x double> %src512) {
 ; SSE-LABEL: 'test_vXf64'
-; SSE-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <2 x double> %src128, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <2 x double> %src128, <2 x double> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <4 x double> %src256, <4 x double> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <2 x double> %src128, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <2 x double> %src128, <2 x double> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <4 x double> %src256, <4 x double> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <4 x double> %src256, <4 x double> %src128_256, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <4 x double> %src256, <4 x double> %src128_256, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <8 x double> %src512, <8 x double> %src128_512, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -32,9 +32,9 @@ define void @test_vXf64(<2 x double> %src128, <4 x double> %src256, <8 x double>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf64'
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <2 x double> %src128, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <2 x double> %src128, <2 x double> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <4 x double> %src256, <4 x double> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <2 x double> %src128, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <2 x double> %src128, <2 x double> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <4 x double> %src256, <4 x double> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <4 x double> %src256, <4 x double> %src128_256, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V256_23 = shufflevector <4 x double> %src256, <4 x double> %src128_256, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <8 x double> %src512, <8 x double> %src128_512, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -46,9 +46,9 @@ define void @test_vXf64(<2 x double> %src128, <4 x double> %src256, <8 x double>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf64'
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <2 x double> %src128, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <2 x double> %src128, <2 x double> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <4 x double> %src256, <4 x double> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <2 x double> %src128, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <2 x double> %src128, <2 x double> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <4 x double> %src256, <4 x double> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <4 x double> %src256, <4 x double> %src128_256, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_23 = shufflevector <4 x double> %src256, <4 x double> %src128_256, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_01 = shufflevector <8 x double> %src512, <8 x double> %src128_512, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -76,9 +76,9 @@ define void @test_vXf64(<2 x double> %src128, <4 x double> %src256, <8 x double>
 
 define void @test_vXi64(<2 x i64> %src128, <4 x i64> %src256, <8 x i64> %src512) {
 ; SSE-LABEL: 'test_vXi64'
-; SSE-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <2 x i64> %src128, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <2 x i64> %src128, <2 x i64> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <4 x i64> %src256, <4 x i64> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <2 x i64> %src128, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <2 x i64> %src128, <2 x i64> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <4 x i64> %src256, <4 x i64> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <4 x i64> %src256, <4 x i64> %src128_256, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <4 x i64> %src256, <4 x i64> %src128_256, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <8 x i64> %src512, <8 x i64> %src128_512, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -90,9 +90,9 @@ define void @test_vXi64(<2 x i64> %src128, <4 x i64> %src256, <8 x i64> %src512)
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi64'
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <2 x i64> %src128, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <2 x i64> %src128, <2 x i64> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <4 x i64> %src256, <4 x i64> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <2 x i64> %src128, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <2 x i64> %src128, <2 x i64> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <4 x i64> %src256, <4 x i64> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <4 x i64> %src256, <4 x i64> %src128_256, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V256_23 = shufflevector <4 x i64> %src256, <4 x i64> %src128_256, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <8 x i64> %src512, <8 x i64> %src128_512, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -104,9 +104,9 @@ define void @test_vXi64(<2 x i64> %src128, <4 x i64> %src256, <8 x i64> %src512)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi64'
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <2 x i64> %src128, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <2 x i64> %src128, <2 x i64> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <4 x i64> %src256, <4 x i64> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <2 x i64> %src128, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <2 x i64> %src128, <2 x i64> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <4 x i64> %src256, <4 x i64> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <4 x i64> %src256, <4 x i64> %src128_256, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_23 = shufflevector <4 x i64> %src256, <4 x i64> %src128_256, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_01 = shufflevector <8 x i64> %src512, <8 x i64> %src128_512, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -134,12 +134,12 @@ define void @test_vXi64(<2 x i64> %src128, <4 x i64> %src256, <8 x i64> %src512)
 
 define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %src256, <16 x float> %src512) {
 ; SSE2-LABEL: 'test_vXf32'
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -165,12 +165,12 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXf32'
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -196,12 +196,12 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXf32'
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -227,12 +227,12 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXf32'
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -258,12 +258,12 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXf32'
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -289,12 +289,12 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf32'
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x float> %src64, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x float> %src64, <2 x float> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x float> %src64, <2 x float> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x float> %src128, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x float> %src128, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x float> %src256, <8 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -355,12 +355,12 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 
 define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256, <16 x i32> %src512) {
 ; SSE2-LABEL: 'test_vXi32'
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -386,12 +386,12 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi32'
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -417,12 +417,12 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi32'
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -448,12 +448,12 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi32'
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -479,12 +479,12 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi32'
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -510,12 +510,12 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi32'
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <2 x i32> %src64, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <2 x i32> %src64, <2 x i32> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <2 x i32> %src64, <2 x i32> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <4 x i32> %src128, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <4 x i32> %src128, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <8 x i32> %src256, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -576,16 +576,16 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 
 define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <16x i16> %src256, <32 x i16> %src512) {
 ; SSE2-LABEL: 'test_vXi16'
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -611,16 +611,16 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi16'
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -646,16 +646,16 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi16'
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -681,16 +681,16 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi16'
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX1-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -716,16 +716,16 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi16'
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX2-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -751,16 +751,16 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi16'
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -786,16 +786,16 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'test_vXi16'
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -821,16 +821,16 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512VMBI-LABEL: 'test_vXi16'
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <2 x i16> %src32, <2 x i16> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <2 x i16> %src32, <2 x i16> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <2 x i16> %src32, <2 x i16> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <2 x i16> %src32, <2 x i16> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <4 x i16> %src64, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <4 x i16> %src64, <4 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <4 x i16> %src64, <4 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <8 x i16> %src128, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <8 x i16> %src128, <8 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <16 x i16> %src256, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> %src32_64, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -896,21 +896,21 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 
 define void @test_vXi8(<2 x i8> %src16, <4 x i8> %src32, <8 x i8> %src64, <16x i8> %src128, <32 x i8> %src256, <64 x i8> %src512) {
 ; SSE2-LABEL: 'test_vXi8'
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE2-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32_01 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32_23 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> %src32_64, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -936,21 +936,21 @@ define void @test_vXi8(<2 x i8> %src16, <4 x i8> %src32, <8 x i8> %src64, <16x i
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi8'
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSSE3-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32_01 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32_23 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> %src32_64, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -976,21 +976,21 @@ define void @test_vXi8(<2 x i8> %src16, <4 x i8> %src32, <8 x i8> %src64, <16x i
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi8'
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; SSE42-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32_01 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32_23 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> %src32_64, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -1016,21 +1016,21 @@ define void @test_vXi8(<2 x i8> %src16, <4 x i8> %src32, <8 x i8> %src64, <16x i
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi8'
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32_01 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32_23 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> %src32_64, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -1056,21 +1056,21 @@ define void @test_vXi8(<2 x i8> %src16, <4 x i8> %src32, <8 x i8> %src64, <16x i
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi8'
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512F-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32_01 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32_23 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> %src32_64, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -1096,21 +1096,21 @@ define void @test_vXi8(<2 x i8> %src16, <4 x i8> %src32, <8 x i8> %src64, <16x i
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'test_vXi8'
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512BW-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32_01 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32_23 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> %src32_64, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -1136,21 +1136,21 @@ define void @test_vXi8(<2 x i8> %src16, <4 x i8> %src32, <8 x i8> %src64, <16x i
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512VMBI-LABEL: 'test_vXi8'
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX512VMBI-NEXT:  Cost Model: Unknown cost for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_32 = shufflevector <2 x i8> %src16, <2 x i8> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_64 = shufflevector <2 x i8> %src16, <2 x i8> undef, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_128 = shufflevector <2 x i8> %src16, <2 x i8> undef, <16 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_256 = shufflevector <2 x i8> %src16, <2 x i8> undef, <32 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src16_512 = shufflevector <2 x i8> %src16, <2 x i8> undef, <64 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_64 = shufflevector <4 x i8> %src32, <4 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_128 = shufflevector <4 x i8> %src32, <4 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_256 = shufflevector <4 x i8> %src32, <4 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src32_512 = shufflevector <4 x i8> %src32, <4 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_128 = shufflevector <8 x i8> %src64, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_256 = shufflevector <8 x i8> %src64, <8 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src64_512 = shufflevector <8 x i8> %src64, <8 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_256 = shufflevector <16 x i8> %src128, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src128_512 = shufflevector <16 x i8> %src128, <16 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %src256_512 = shufflevector <32 x i8> %src256, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32_01 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32_23 = shufflevector <4 x i8> %src32, <4 x i8> %src16_32, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; AVX512VMBI-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> %src32_64, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>