From 543236232c79221c4da261246e49888844697539 Mon Sep 17 00:00:00 2001
From: David Green <david.green@arm.com>
Date: Mon, 14 Oct 2019 15:19:33 +0000
Subject: [PATCH] [ARM] Selection for MVE VMOVN

The adds both VMOVNt and VMOVNb instruction selection from the appropriate
shuffles. We detect shuffle masks of the form:
0, N, 2, N+2, 4, N+4, ...
or
0, N+1, 2, N+3, 4, N+5, ...
ISel will also try the opposite patterns, with inputs reversed. These are
selected to VMOVNt and VMOVNb respectively.

Differential Revision: https://reviews.llvm.org/D68283

llvm-svn: 374781
---
 llvm/lib/Target/ARM/ARMISelLowering.cpp |  35 ++++
 llvm/lib/Target/ARM/ARMISelLowering.h   |   1 +
 llvm/lib/Target/ARM/ARMInstrMVE.td      |  12 ++
 llvm/test/CodeGen/Thumb2/mve-vmovn.ll   | 318 ++------------------------------
 4 files changed, 66 insertions(+), 300 deletions(-)

diff --git a/llvm/lib/Target/ARM/ARMISelLowering.cpp b/llvm/lib/Target/ARM/ARMISelLowering.cpp
index d8b2c55..71d53a3 100644
--- a/llvm/lib/Target/ARM/ARMISelLowering.cpp
+++ b/llvm/lib/Target/ARM/ARMISelLowering.cpp
@@ -1604,6 +1604,7 @@ const char *ARMTargetLowering::getTargetNodeName(unsigned Opcode) const {
   case ARMISD::VTRN:          return "ARMISD::VTRN";
   case ARMISD::VTBL1:         return "ARMISD::VTBL1";
   case ARMISD::VTBL2:         return "ARMISD::VTBL2";
+  case ARMISD::VMOVN:         return "ARMISD::VMOVN";
   case ARMISD::VMULLs:        return "ARMISD::VMULLs";
   case ARMISD::VMULLu:        return "ARMISD::VMULLu";
   case ARMISD::UMAAL:         return "ARMISD::UMAAL";
@@ -6891,6 +6892,29 @@ static bool isReverseMask(ArrayRef<int> M, EVT VT) {
   return true;
 }
 
+static bool isVMOVNMask(ArrayRef<int> M, EVT VT, bool Top) {
+  unsigned NumElts = VT.getVectorNumElements();
+  // Make sure the mask has the right size.
+  if (NumElts != M.size() || (VT != MVT::v8i16 && VT != MVT::v16i8))
+      return false;
+
+  // If Top
+  //   Look for <0, N, 2, N+2, 4, N+4, ..>.
+  //   This inserts Input2 into Input1
+  // else if not Top
+  //   Look for <0, N+1, 2, N+3, 4, N+5, ..>
+  //   This inserts Input1 into Input2
+  unsigned Offset = Top ? 0 : 1;
+  for (unsigned i = 0; i < NumElts; i+=2) {
+    if (M[i] >= 0 && M[i] != (int)i)
+      return false;
+    if (M[i+1] >= 0 && M[i+1] != (int)(NumElts + i + Offset))
+      return false;
+  }
+
+  return true;
+}
+
 // If N is an integer constant that can be moved into a register in one
 // instruction, return an SDValue of such a constant (will become a MOV
 // instruction).  Otherwise return null.
@@ -7485,6 +7509,9 @@ bool ARMTargetLowering::isShuffleMaskLegal(ArrayRef<int> M, EVT VT) const {
   else if (Subtarget->hasNEON() && (VT == MVT::v8i16 || VT == MVT::v16i8) &&
            isReverseMask(M, VT))
     return true;
+  else if (Subtarget->hasMVEIntegerOps() &&
+           (isVMOVNMask(M, VT, 0) || isVMOVNMask(M, VT, 1)))
+    return true;
   else
     return false;
 }
@@ -7760,6 +7787,14 @@ static SDValue LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG,
             .getValue(WhichResult);
       }
     }
+    if (ST->hasMVEIntegerOps()) {
+      if (isVMOVNMask(ShuffleMask, VT, 0))
+        return DAG.getNode(ARMISD::VMOVN, dl, VT, V2, V1,
+                           DAG.getConstant(0, dl, MVT::i32));
+      if (isVMOVNMask(ShuffleMask, VT, 1))
+        return DAG.getNode(ARMISD::VMOVN, dl, VT, V1, V2,
+                           DAG.getConstant(1, dl, MVT::i32));
+    }
 
     // Also check for these shuffles through CONCAT_VECTORS: we canonicalize
     // shuffles that produce a result larger than their operands with:
diff --git a/llvm/lib/Target/ARM/ARMISelLowering.h b/llvm/lib/Target/ARM/ARMISelLowering.h
index a89ef25..98e0684 100644
--- a/llvm/lib/Target/ARM/ARMISelLowering.h
+++ b/llvm/lib/Target/ARM/ARMISelLowering.h
@@ -197,6 +197,7 @@ class VectorType;
       VTRN,         // transpose
       VTBL1,        // 1-register shuffle with mask
       VTBL2,        // 2-register shuffle with mask
+      VMOVN,        // MVE vmovn
 
       // Vector multiply long:
       VMULLs,       // ...signed
diff --git a/llvm/lib/Target/ARM/ARMInstrMVE.td b/llvm/lib/Target/ARM/ARMInstrMVE.td
index 80b45ce..dad9d51 100644
--- a/llvm/lib/Target/ARM/ARMInstrMVE.td
+++ b/llvm/lib/Target/ARM/ARMInstrMVE.td
@@ -3501,6 +3501,18 @@ defm MVE_VQMOVNu32  : MVE_VxMOVxN_halves<"vqmovn",  "u32", 0b1, 0b1, 0b01>;
 defm MVE_VQMOVUNs16 : MVE_VxMOVxN_halves<"vqmovun", "s16", 0b0, 0b0, 0b00>;
 defm MVE_VQMOVUNs32 : MVE_VxMOVxN_halves<"vqmovun", "s32", 0b0, 0b0, 0b01>;
 
+def MVEvmovn       : SDNode<"ARMISD::VMOVN", SDTARMVEXT>;
+let Predicates = [HasMVEInt] in {
+  def : Pat<(v8i16 (MVEvmovn (v8i16 MQPR:$Qd_src), (v8i16 MQPR:$Qm), (i32 0))),
+            (v8i16 (MVE_VMOVNi32bh (v8i16 MQPR:$Qd_src), (v8i16 MQPR:$Qm)))>;
+  def : Pat<(v8i16 (MVEvmovn (v8i16 MQPR:$Qd_src), (v8i16 MQPR:$Qm), (i32 1))),
+            (v8i16 (MVE_VMOVNi32th (v8i16 MQPR:$Qd_src), (v8i16 MQPR:$Qm)))>;
+  def : Pat<(v16i8 (MVEvmovn (v16i8 MQPR:$Qd_src), (v16i8 MQPR:$Qm), (i32 0))),
+            (v16i8 (MVE_VMOVNi16bh (v16i8 MQPR:$Qd_src), (v16i8 MQPR:$Qm)))>;
+  def : Pat<(v16i8 (MVEvmovn (v16i8 MQPR:$Qd_src), (v16i8 MQPR:$Qm), (i32 1))),
+            (v16i8 (MVE_VMOVNi16th (v16i8 MQPR:$Qd_src), (v16i8 MQPR:$Qm)))>;
+}
+
 class MVE_VCVT_ff<string iname, string suffix, bit op, bit T,
                   list<dag> pattern=[]>
   : MVE_qDest_qSrc<iname, suffix, (outs MQPR:$Qd), (ins MQPR:$Qd_src, MQPR:$Qm),
diff --git a/llvm/test/CodeGen/Thumb2/mve-vmovn.ll b/llvm/test/CodeGen/Thumb2/mve-vmovn.ll
index 642c5ff..20f790d 100644
--- a/llvm/test/CodeGen/Thumb2/mve-vmovn.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vmovn.ll
@@ -4,23 +4,7 @@
 define arm_aapcs_vfpcc <8 x i16> @vmovn32_trunc1(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-LABEL: vmovn32_trunc1:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov r0, s8
-; CHECK-NEXT:    vmov.16 q0[0], r0
-; CHECK-NEXT:    vmov r0, s4
-; CHECK-NEXT:    vmov.16 q0[1], r0
-; CHECK-NEXT:    vmov r0, s9
-; CHECK-NEXT:    vmov.16 q0[2], r0
-; CHECK-NEXT:    vmov r0, s5
-; CHECK-NEXT:    vmov.16 q0[3], r0
-; CHECK-NEXT:    vmov r0, s10
-; CHECK-NEXT:    vmov.16 q0[4], r0
-; CHECK-NEXT:    vmov r0, s6
-; CHECK-NEXT:    vmov.16 q0[5], r0
-; CHECK-NEXT:    vmov r0, s11
-; CHECK-NEXT:    vmov.16 q0[6], r0
-; CHECK-NEXT:    vmov r0, s7
-; CHECK-NEXT:    vmov.16 q0[7], r0
+; CHECK-NEXT:    vmovnt.i32 q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %strided.vec = shufflevector <4 x i32> %src1, <4 x i32> %src2, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
@@ -31,23 +15,8 @@ entry:
 define arm_aapcs_vfpcc <8 x i16> @vmovn32_trunc2(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-LABEL: vmovn32_trunc2:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov r0, s4
-; CHECK-NEXT:    vmov.16 q0[0], r0
-; CHECK-NEXT:    vmov r0, s8
-; CHECK-NEXT:    vmov.16 q0[1], r0
-; CHECK-NEXT:    vmov r0, s5
-; CHECK-NEXT:    vmov.16 q0[2], r0
-; CHECK-NEXT:    vmov r0, s9
-; CHECK-NEXT:    vmov.16 q0[3], r0
-; CHECK-NEXT:    vmov r0, s6
-; CHECK-NEXT:    vmov.16 q0[4], r0
-; CHECK-NEXT:    vmov r0, s10
-; CHECK-NEXT:    vmov.16 q0[5], r0
-; CHECK-NEXT:    vmov r0, s7
-; CHECK-NEXT:    vmov.16 q0[6], r0
-; CHECK-NEXT:    vmov r0, s11
-; CHECK-NEXT:    vmov.16 q0[7], r0
+; CHECK-NEXT:    vmovnt.i32 q1, q0
+; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %strided.vec = shufflevector <4 x i32> %src1, <4 x i32> %src2, <8 x i32> <i32 4, i32 0, i32 5, i32 1, i32 6, i32 2, i32 7, i32 3>
@@ -58,39 +27,7 @@ entry:
 define arm_aapcs_vfpcc <16 x i8> @vmovn16_trunc1(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-LABEL: vmovn16_trunc1:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u16 r0, q0[0]
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.8 q0[0], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[0]
-; CHECK-NEXT:    vmov.8 q0[1], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[1]
-; CHECK-NEXT:    vmov.8 q0[2], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[1]
-; CHECK-NEXT:    vmov.8 q0[3], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[2]
-; CHECK-NEXT:    vmov.8 q0[4], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[2]
-; CHECK-NEXT:    vmov.8 q0[5], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[3]
-; CHECK-NEXT:    vmov.8 q0[6], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[3]
-; CHECK-NEXT:    vmov.8 q0[7], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[4]
-; CHECK-NEXT:    vmov.8 q0[8], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[4]
-; CHECK-NEXT:    vmov.8 q0[9], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[5]
-; CHECK-NEXT:    vmov.8 q0[10], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[5]
-; CHECK-NEXT:    vmov.8 q0[11], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[6]
-; CHECK-NEXT:    vmov.8 q0[12], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[6]
-; CHECK-NEXT:    vmov.8 q0[13], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[7]
-; CHECK-NEXT:    vmov.8 q0[14], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[7]
-; CHECK-NEXT:    vmov.8 q0[15], r0
+; CHECK-NEXT:    vmovnt.i16 q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %strided.vec = shufflevector <8 x i16> %src1, <8 x i16> %src2, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
@@ -101,39 +38,8 @@ entry:
 define arm_aapcs_vfpcc <16 x i8> @vmovn16_trunc2(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-LABEL: vmovn16_trunc2:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.u16 r0, q1[0]
-; CHECK-NEXT:    vmov.8 q0[0], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[0]
-; CHECK-NEXT:    vmov.8 q0[1], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[1]
-; CHECK-NEXT:    vmov.8 q0[2], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[1]
-; CHECK-NEXT:    vmov.8 q0[3], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[2]
-; CHECK-NEXT:    vmov.8 q0[4], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[2]
-; CHECK-NEXT:    vmov.8 q0[5], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[3]
-; CHECK-NEXT:    vmov.8 q0[6], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[3]
-; CHECK-NEXT:    vmov.8 q0[7], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[4]
-; CHECK-NEXT:    vmov.8 q0[8], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[4]
-; CHECK-NEXT:    vmov.8 q0[9], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[5]
-; CHECK-NEXT:    vmov.8 q0[10], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[5]
-; CHECK-NEXT:    vmov.8 q0[11], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[6]
-; CHECK-NEXT:    vmov.8 q0[12], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[6]
-; CHECK-NEXT:    vmov.8 q0[13], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[7]
-; CHECK-NEXT:    vmov.8 q0[14], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[7]
-; CHECK-NEXT:    vmov.8 q0[15], r0
+; CHECK-NEXT:    vmovnt.i16 q1, q0
+; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %strided.vec = shufflevector <8 x i16> %src1, <8 x i16> %src2, <16 x i32> <i32 8, i32 0, i32 9, i32 1, i32 10, i32 2, i32 11, i32 3, i32 12, i32 4, i32 13, i32 5, i32 14, i32 6, i32 15, i32 7>
@@ -297,23 +203,7 @@ entry:
 define arm_aapcs_vfpcc <8 x i16> @vmovn16_t1(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-LABEL: vmovn16_t1:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u16 r0, q0[0]
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.16 q0[0], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[0]
-; CHECK-NEXT:    vmov.16 q0[1], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[2]
-; CHECK-NEXT:    vmov.16 q0[2], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q0[3], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[4]
-; CHECK-NEXT:    vmov.16 q0[4], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q0[5], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[6]
-; CHECK-NEXT:    vmov.16 q0[6], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q0[7], r0
+; CHECK-NEXT:    vmovnt.i32 q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
@@ -323,23 +213,8 @@ entry:
 define arm_aapcs_vfpcc <8 x i16> @vmovn16_t2(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-LABEL: vmovn16_t2:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.u16 r0, q1[0]
-; CHECK-NEXT:    vmov.16 q0[0], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[0]
-; CHECK-NEXT:    vmov.16 q0[1], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q0[2], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[2]
-; CHECK-NEXT:    vmov.16 q0[3], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q0[4], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[4]
-; CHECK-NEXT:    vmov.16 q0[5], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q0[6], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[6]
-; CHECK-NEXT:    vmov.16 q0[7], r0
+; CHECK-NEXT:    vmovnt.i32 q1, q0
+; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 8, i32 0, i32 10, i32 2, i32 12, i32 4, i32 14, i32 6>
@@ -349,23 +224,8 @@ entry:
 define arm_aapcs_vfpcc <8 x i16> @vmovn16_b1(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-LABEL: vmovn16_b1:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u16 r0, q0[0]
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.16 q0[0], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[1]
-; CHECK-NEXT:    vmov.16 q0[1], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[2]
-; CHECK-NEXT:    vmov.16 q0[2], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[3]
-; CHECK-NEXT:    vmov.16 q0[3], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[4]
-; CHECK-NEXT:    vmov.16 q0[4], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[5]
-; CHECK-NEXT:    vmov.16 q0[5], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[6]
-; CHECK-NEXT:    vmov.16 q0[6], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[7]
-; CHECK-NEXT:    vmov.16 q0[7], r0
+; CHECK-NEXT:    vmovnb.i32 q1, q0
+; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
@@ -427,23 +287,7 @@ entry:
 define arm_aapcs_vfpcc <8 x i16> @vmovn16_b4(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-LABEL: vmovn16_b4:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.u16 r0, q1[0]
-; CHECK-NEXT:    vmov.16 q0[0], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[1]
-; CHECK-NEXT:    vmov.16 q0[1], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q0[2], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[3]
-; CHECK-NEXT:    vmov.16 q0[3], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q0[4], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[5]
-; CHECK-NEXT:    vmov.16 q0[5], r0
-; CHECK-NEXT:    vmov.u16 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q0[6], r0
-; CHECK-NEXT:    vmov.u16 r0, q2[7]
-; CHECK-NEXT:    vmov.16 q0[7], r0
+; CHECK-NEXT:    vmovnb.i32 q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
@@ -454,39 +298,7 @@ entry:
 define arm_aapcs_vfpcc <16 x i8> @vmovn8_b1(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-LABEL: vmovn8_b1:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.8 q0[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vmov.8 q0[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[2]
-; CHECK-NEXT:    vmov.8 q0[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.8 q0[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[4]
-; CHECK-NEXT:    vmov.8 q0[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.8 q0[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[6]
-; CHECK-NEXT:    vmov.8 q0[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.8 q0[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[8]
-; CHECK-NEXT:    vmov.8 q0[8], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmov.8 q0[9], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[10]
-; CHECK-NEXT:    vmov.8 q0[10], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.8 q0[11], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[12]
-; CHECK-NEXT:    vmov.8 q0[12], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.8 q0[13], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[14]
-; CHECK-NEXT:    vmov.8 q0[14], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.8 q0[15], r0
+; CHECK-NEXT:    vmovnt.i16 q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
@@ -496,39 +308,8 @@ entry:
 define arm_aapcs_vfpcc <16 x i8> @vmovn8_b2(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-LABEL: vmovn8_b2:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vmov.8 q0[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[0]
-; CHECK-NEXT:    vmov.8 q0[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.8 q0[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[2]
-; CHECK-NEXT:    vmov.8 q0[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.8 q0[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[4]
-; CHECK-NEXT:    vmov.8 q0[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.8 q0[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[6]
-; CHECK-NEXT:    vmov.8 q0[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmov.8 q0[8], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[8]
-; CHECK-NEXT:    vmov.8 q0[9], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.8 q0[10], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[10]
-; CHECK-NEXT:    vmov.8 q0[11], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.8 q0[12], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[12]
-; CHECK-NEXT:    vmov.8 q0[13], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.8 q0[14], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[14]
-; CHECK-NEXT:    vmov.8 q0[15], r0
+; CHECK-NEXT:    vmovnt.i16 q1, q0
+; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 16, i32 0, i32 18, i32 2, i32 20, i32 4, i32 22, i32 6, i32 24, i32 8, i32 26, i32 10, i32 28, i32 12, i32 30, i32 14>
@@ -538,39 +319,8 @@ entry:
 define arm_aapcs_vfpcc <16 x i8> @vmovn8_t1(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-LABEL: vmovn8_t1:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.8 q0[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[1]
-; CHECK-NEXT:    vmov.8 q0[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[2]
-; CHECK-NEXT:    vmov.8 q0[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[3]
-; CHECK-NEXT:    vmov.8 q0[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[4]
-; CHECK-NEXT:    vmov.8 q0[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[5]
-; CHECK-NEXT:    vmov.8 q0[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[6]
-; CHECK-NEXT:    vmov.8 q0[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[7]
-; CHECK-NEXT:    vmov.8 q0[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[8]
-; CHECK-NEXT:    vmov.8 q0[8], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[9]
-; CHECK-NEXT:    vmov.8 q0[9], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[10]
-; CHECK-NEXT:    vmov.8 q0[10], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[11]
-; CHECK-NEXT:    vmov.8 q0[11], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[12]
-; CHECK-NEXT:    vmov.8 q0[12], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[13]
-; CHECK-NEXT:    vmov.8 q0[13], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[14]
-; CHECK-NEXT:    vmov.8 q0[14], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[15]
-; CHECK-NEXT:    vmov.8 q0[15], r0
+; CHECK-NEXT:    vmovnb.i16 q1, q0
+; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
@@ -664,39 +414,7 @@ entry:
 define arm_aapcs_vfpcc <16 x i8> @vmovn8_t4(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-LABEL: vmovn8_t4:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov q2, q0
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vmov.8 q0[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[1]
-; CHECK-NEXT:    vmov.8 q0[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.8 q0[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[3]
-; CHECK-NEXT:    vmov.8 q0[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.8 q0[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[5]
-; CHECK-NEXT:    vmov.8 q0[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.8 q0[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[7]
-; CHECK-NEXT:    vmov.8 q0[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmov.8 q0[8], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[9]
-; CHECK-NEXT:    vmov.8 q0[9], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.8 q0[10], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[11]
-; CHECK-NEXT:    vmov.8 q0[11], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.8 q0[12], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[13]
-; CHECK-NEXT:    vmov.8 q0[13], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.8 q0[14], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[15]
-; CHECK-NEXT:    vmov.8 q0[15], r0
+; CHECK-NEXT:    vmovnb.i16 q0, q1
 ; CHECK-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
-- 
2.7.4