[Cost][X86] Add more missing vector truncation costs

The AVX512 cases still need some work to correct recognise the PMOV truncation cases.

llvm-svn: 372514

diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index 9b4a386..10ece8a 100644 (file)
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -1480,7 +1480,10 @@ int X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src,
     { ISD::TRUNCATE,    MVT::v4i8,  MVT::v4i64,  4 },
     { ISD::TRUNCATE,    MVT::v4i16, MVT::v4i64,  4 },
     { ISD::TRUNCATE,    MVT::v4i32, MVT::v4i64,  4 },
+    { ISD::TRUNCATE,    MVT::v8i8,  MVT::v8i64, 11 },
+    { ISD::TRUNCATE,    MVT::v8i16, MVT::v8i64,  9 },
     { ISD::TRUNCATE,    MVT::v8i32, MVT::v8i64,  9 },
+    { ISD::TRUNCATE,    MVT::v16i8, MVT::v16i64, 11 },
 
     { ISD::SINT_TO_FP,  MVT::v4f32, MVT::v4i1,  3 },
     { ISD::SINT_TO_FP,  MVT::v4f64, MVT::v4i1,  3 },
@@ -1629,9 +1632,12 @@ int X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src,
     { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i32,  3 },
     { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i32,  5 },
 
+    { ISD::TRUNCATE,    MVT::v2i8,   MVT::v2i16,  2 }, // PAND+PACKUSWB
     { ISD::TRUNCATE,    MVT::v4i8,   MVT::v4i16,  4 },
     { ISD::TRUNCATE,    MVT::v8i8,   MVT::v8i16,  2 },
     { ISD::TRUNCATE,    MVT::v16i8,  MVT::v16i16, 3 },
+    { ISD::TRUNCATE,    MVT::v2i8,   MVT::v2i32,  3 }, // PAND+3*PACKUSWB
+    { ISD::TRUNCATE,    MVT::v2i16,  MVT::v2i32,  1 },
     { ISD::TRUNCATE,    MVT::v4i8,   MVT::v4i32,  3 },
     { ISD::TRUNCATE,    MVT::v4i16,  MVT::v4i32,  3 },
     { ISD::TRUNCATE,    MVT::v8i8,   MVT::v8i32,  4 },

diff --git a/llvm/test/Analysis/CostModel/X86/trunc.ll b/llvm/test/Analysis/CostModel/X86/trunc.ll
index 31b9933..37d18c3 100644 (file)
--- a/llvm/test/Analysis/CostModel/X86/trunc.ll
+++ b/llvm/test/Analysis/CostModel/X86/trunc.ll
@@ -61,7 +61,7 @@ define i32 @trunc_vXi16() {
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
@@ -74,7 +74,7 @@ define i32 @trunc_vXi16() {
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
@@ -87,7 +87,7 @@ define i32 @trunc_vXi16() {
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
@@ -97,10 +97,10 @@ define i32 @trunc_vXi16() {
 ; AVX1-LABEL: 'trunc_vXi16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i16>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i16>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
@@ -110,10 +110,10 @@ define i32 @trunc_vXi16() {
 ; AVX2-LABEL: 'trunc_vXi16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i16>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
@@ -126,7 +126,7 @@ define i32 @trunc_vXi16() {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
@@ -139,7 +139,7 @@ define i32 @trunc_vXi16() {
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
@@ -149,10 +149,10 @@ define i32 @trunc_vXi16() {
 ; BTVER2-LABEL: 'trunc_vXi16'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i16>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i16>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i16>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
@@ -181,13 +181,13 @@ define i32 @trunc_vXi8() {
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 159 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
@@ -202,13 +202,13 @@ define i32 @trunc_vXi8() {
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 159 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
@@ -223,13 +223,13 @@ define i32 @trunc_vXi8() {
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
@@ -240,17 +240,17 @@ define i32 @trunc_vXi8() {
 ; AVX1-LABEL: 'trunc_vXi8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
@@ -261,17 +261,17 @@ define i32 @trunc_vXi8() {
 ; AVX2-LABEL: 'trunc_vXi8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
@@ -282,17 +282,17 @@ define i32 @trunc_vXi8() {
 ; AVX512F-LABEL: 'trunc_vXi8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
@@ -303,17 +303,17 @@ define i32 @trunc_vXi8() {
 ; AVX512BW-LABEL: 'trunc_vXi8'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
@@ -324,17 +324,17 @@ define i32 @trunc_vXi8() {
 ; BTVER2-LABEL: 'trunc_vXi8'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16i64 = trunc <16 x i64> undef to <16 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32i64 = trunc <32 x i64> undef to <32 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V64i64 = trunc <64 x i64> undef to <64 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32i32 = trunc <32 x i32> undef to <32 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64i32 = trunc <64 x i32> undef to <64 x i8>
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>