From 186a7f81e813a59b68fadfe7ea90c348bb760c56 Mon Sep 17 00:00:00 2001
From: David Green <david.green@arm.com>
Date: Sun, 9 Aug 2020 11:09:49 +0100
Subject: [PATCH] [ARM] Add VADDV and VMLAV patterns for v16i16

This adds patterns for v16i16's vecreduce, using all the existing code
to go via an i32 VADDV/VMLAV and truncating the result.

Differential Revision: https://reviews.llvm.org/D85452
---
 llvm/lib/Target/ARM/ARMISelLowering.cpp           |  24 ++
 llvm/test/CodeGen/Thumb2/mve-vecreduce-add.ll     | 148 +------
 llvm/test/CodeGen/Thumb2/mve-vecreduce-addpred.ll | 312 +--------------
 llvm/test/CodeGen/Thumb2/mve-vecreduce-loops.ll   | 252 +-----------
 llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll     | 276 +------------
 llvm/test/CodeGen/Thumb2/mve-vecreduce-mlapred.ll | 460 +---------------------
 6 files changed, 67 insertions(+), 1405 deletions(-)

diff --git a/llvm/lib/Target/ARM/ARMISelLowering.cpp b/llvm/lib/Target/ARM/ARMISelLowering.cpp
index dd53b10..42c0db7 100644
--- a/llvm/lib/Target/ARM/ARMISelLowering.cpp
+++ b/llvm/lib/Target/ARM/ARMISelLowering.cpp
@@ -14808,6 +14808,12 @@ static SDValue PerformVECREDUCE_ADDCombine(SDNode *N, SelectionDAG &DAG,
     return Create64bitNode(ARMISD::VADDLVs, {A});
   if (SDValue A = IsVADDV(MVT::i64, ISD::ZERO_EXTEND, {MVT::v4i32}))
     return Create64bitNode(ARMISD::VADDLVu, {A});
+  if (SDValue A = IsVADDV(MVT::i16, ISD::SIGN_EXTEND, {MVT::v16i8}))
+    return DAG.getNode(ISD::TRUNCATE, dl, ResVT,
+                       DAG.getNode(ARMISD::VADDVs, dl, MVT::i32, A));
+  if (SDValue A = IsVADDV(MVT::i16, ISD::ZERO_EXTEND, {MVT::v16i8}))
+    return DAG.getNode(ISD::TRUNCATE, dl, ResVT,
+                       DAG.getNode(ARMISD::VADDVu, dl, MVT::i32, A));
 
   SDValue Mask;
   if (SDValue A = IsPredVADDV(MVT::i32, ISD::SIGN_EXTEND, {MVT::v8i16, MVT::v16i8}, Mask))
@@ -14818,6 +14824,12 @@ static SDValue PerformVECREDUCE_ADDCombine(SDNode *N, SelectionDAG &DAG,
     return Create64bitNode(ARMISD::VADDLVps, {A, Mask});
   if (SDValue A = IsPredVADDV(MVT::i64, ISD::ZERO_EXTEND, {MVT::v4i32}, Mask))
     return Create64bitNode(ARMISD::VADDLVpu, {A, Mask});
+  if (SDValue A = IsPredVADDV(MVT::i16, ISD::SIGN_EXTEND, {MVT::v16i8}, Mask))
+    return DAG.getNode(ISD::TRUNCATE, dl, ResVT,
+                       DAG.getNode(ARMISD::VADDVps, dl, MVT::i32, A, Mask));
+  if (SDValue A = IsPredVADDV(MVT::i16, ISD::ZERO_EXTEND, {MVT::v16i8}, Mask))
+    return DAG.getNode(ISD::TRUNCATE, dl, ResVT,
+                       DAG.getNode(ARMISD::VADDVpu, dl, MVT::i32, A, Mask));
 
   SDValue A, B;
   if (IsVMLAV(MVT::i32, ISD::SIGN_EXTEND, {MVT::v8i16, MVT::v16i8}, A, B))
@@ -14828,6 +14840,12 @@ static SDValue PerformVECREDUCE_ADDCombine(SDNode *N, SelectionDAG &DAG,
     return Create64bitNode(ARMISD::VMLALVs, {A, B});
   if (IsVMLAV(MVT::i64, ISD::ZERO_EXTEND, {MVT::v8i16, MVT::v4i32}, A, B))
     return Create64bitNode(ARMISD::VMLALVu, {A, B});
+  if (IsVMLAV(MVT::i16, ISD::SIGN_EXTEND, {MVT::v16i8}, A, B))
+    return DAG.getNode(ISD::TRUNCATE, dl, ResVT,
+                       DAG.getNode(ARMISD::VMLAVs, dl, MVT::i32, A, B));
+  if (IsVMLAV(MVT::i16, ISD::ZERO_EXTEND, {MVT::v16i8}, A, B))
+    return DAG.getNode(ISD::TRUNCATE, dl, ResVT,
+                       DAG.getNode(ARMISD::VMLAVu, dl, MVT::i32, A, B));
 
   if (IsPredVMLAV(MVT::i32, ISD::SIGN_EXTEND, {MVT::v8i16, MVT::v16i8}, A, B, Mask))
     return DAG.getNode(ARMISD::VMLAVps, dl, ResVT, A, B, Mask);
@@ -14837,6 +14855,12 @@ static SDValue PerformVECREDUCE_ADDCombine(SDNode *N, SelectionDAG &DAG,
     return Create64bitNode(ARMISD::VMLALVps, {A, B, Mask});
   if (IsPredVMLAV(MVT::i64, ISD::ZERO_EXTEND, {MVT::v8i16, MVT::v4i32}, A, B, Mask))
     return Create64bitNode(ARMISD::VMLALVpu, {A, B, Mask});
+  if (IsPredVMLAV(MVT::i16, ISD::SIGN_EXTEND, {MVT::v16i8}, A, B, Mask))
+    return DAG.getNode(ISD::TRUNCATE, dl, ResVT,
+                       DAG.getNode(ARMISD::VMLAVps, dl, MVT::i32, A, B, Mask));
+  if (IsPredVMLAV(MVT::i16, ISD::ZERO_EXTEND, {MVT::v16i8}, A, B, Mask))
+    return DAG.getNode(ISD::TRUNCATE, dl, ResVT,
+                       DAG.getNode(ARMISD::VMLAVpu, dl, MVT::i32, A, B, Mask));
   return SDValue();
 }
 
diff --git a/llvm/test/CodeGen/Thumb2/mve-vecreduce-add.ll b/llvm/test/CodeGen/Thumb2/mve-vecreduce-add.ll
index 2f483b3..35eecab 100644
--- a/llvm/test/CodeGen/Thumb2/mve-vecreduce-add.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vecreduce-add.ll
@@ -333,42 +333,7 @@ entry:
 define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_zext(<16 x i8> %x) {
 ; CHECK-LABEL: add_v16i8_v16i16_zext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r0, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[1]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[2]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[3]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[4]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[5]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[6]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[7]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmovlb.u8 q1, q1
-; CHECK-NEXT:    vmovlb.u8 q0, q2
-; CHECK-NEXT:    vadd.i16 q0, q0, q1
-; CHECK-NEXT:    vaddv.u16 r0, q0
+; CHECK-NEXT:    vaddv.u8 r0, q0
 ; CHECK-NEXT:    uxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -380,42 +345,7 @@ entry:
 define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_sext(<16 x i8> %x) {
 ; CHECK-LABEL: add_v16i8_v16i16_sext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r0, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[1]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[2]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[3]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[4]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[5]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[6]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[7]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmovlb.s8 q1, q1
-; CHECK-NEXT:    vmovlb.s8 q0, q2
-; CHECK-NEXT:    vadd.i16 q0, q0, q1
-; CHECK-NEXT:    vaddv.u16 r0, q0
+; CHECK-NEXT:    vaddv.s8 r0, q0
 ; CHECK-NEXT:    sxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -1120,42 +1050,7 @@ entry:
 define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_acc_zext(<16 x i8> %x, i16 %a) {
 ; CHECK-LABEL: add_v16i8_v16i16_acc_zext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r1, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[0]
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[1]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[2]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[3]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[4]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[5]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[6]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[7]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmovlb.u8 q1, q1
-; CHECK-NEXT:    vmovlb.u8 q0, q2
-; CHECK-NEXT:    vadd.i16 q0, q0, q1
-; CHECK-NEXT:    vaddva.u16 r0, q0
+; CHECK-NEXT:    vaddva.u8 r0, q0
 ; CHECK-NEXT:    uxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -1168,42 +1063,7 @@ entry:
 define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_acc_sext(<16 x i8> %x, i16 %a) {
 ; CHECK-LABEL: add_v16i8_v16i16_acc_sext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r1, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[0]
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[1]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[2]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[3]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[4]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[5]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[6]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[7]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmovlb.s8 q1, q1
-; CHECK-NEXT:    vmovlb.s8 q0, q2
-; CHECK-NEXT:    vadd.i16 q0, q0, q1
-; CHECK-NEXT:    vaddva.u16 r0, q0
+; CHECK-NEXT:    vaddva.s8 r0, q0
 ; CHECK-NEXT:    sxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
diff --git a/llvm/test/CodeGen/Thumb2/mve-vecreduce-addpred.ll b/llvm/test/CodeGen/Thumb2/mve-vecreduce-addpred.ll
index eb47d1d..0f3aacf 100644
--- a/llvm/test/CodeGen/Thumb2/mve-vecreduce-addpred.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vecreduce-addpred.ll
@@ -607,82 +607,8 @@ entry:
 define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_zext(<16 x i8> %x, <16 x i8> %b) {
 ; CHECK-LABEL: add_v16i8_v16i16_zext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vcmp.i8 eq, q1, zr
-; CHECK-NEXT:    vmov.i8 q1, #0x0
-; CHECK-NEXT:    vmov.i8 q2, #0xff
-; CHECK-NEXT:    vmov.i32 q3, #0x0
-; CHECK-NEXT:    vpsel q1, q2, q1
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[1]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[3]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[5]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[7]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vcmp.i16 ne, q2, zr
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[1]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[2]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[3]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[4]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[5]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[6]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[7]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmovlb.u8 q2, q2
-; CHECK-NEXT:    vpsel q2, q2, q3
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[9]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[11]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[13]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[15]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r0
-; CHECK-NEXT:    vmovlb.u8 q0, q1
-; CHECK-NEXT:    vpt.i16 ne, q3, zr
-; CHECK-NEXT:    vaddt.i16 q2, q2, q0
-; CHECK-NEXT:    vaddv.u16 r0, q2
+; CHECK-NEXT:    vpt.i8 eq, q1, zr
+; CHECK-NEXT:    vaddvt.u8 r0, q0
 ; CHECK-NEXT:    uxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -696,82 +622,8 @@ entry:
 define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_sext(<16 x i8> %x, <16 x i8> %b) {
 ; CHECK-LABEL: add_v16i8_v16i16_sext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vcmp.i8 eq, q1, zr
-; CHECK-NEXT:    vmov.i8 q1, #0x0
-; CHECK-NEXT:    vmov.i8 q2, #0xff
-; CHECK-NEXT:    vmov.i32 q3, #0x0
-; CHECK-NEXT:    vpsel q1, q2, q1
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[1]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[3]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[5]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[7]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vcmp.i16 ne, q2, zr
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[1]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[2]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[3]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[4]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[5]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[6]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[7]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmovlb.s8 q2, q2
-; CHECK-NEXT:    vpsel q2, q2, q3
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[9]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[11]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[13]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[15]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r0
-; CHECK-NEXT:    vmovlb.s8 q0, q1
-; CHECK-NEXT:    vpt.i16 ne, q3, zr
-; CHECK-NEXT:    vaddt.i16 q2, q2, q0
-; CHECK-NEXT:    vaddv.u16 r0, q2
+; CHECK-NEXT:    vpt.i8 eq, q1, zr
+; CHECK-NEXT:    vaddvt.s8 r0, q0
 ; CHECK-NEXT:    sxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -2170,82 +2022,8 @@ entry:
 define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_acc_zext(<16 x i8> %x, <16 x i8> %b, i16 %a) {
 ; CHECK-LABEL: add_v16i8_v16i16_acc_zext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vcmp.i8 eq, q1, zr
-; CHECK-NEXT:    vmov.i8 q1, #0x0
-; CHECK-NEXT:    vmov.i8 q2, #0xff
-; CHECK-NEXT:    vmov.i32 q3, #0x0
-; CHECK-NEXT:    vpsel q1, q2, q1
-; CHECK-NEXT:    vmov.u8 r1, q1[0]
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[1]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[2]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[3]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[4]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[5]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[6]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[7]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[0]
-; CHECK-NEXT:    vcmp.i16 ne, q2, zr
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[1]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[2]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[3]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[4]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[5]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[6]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[7]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[8]
-; CHECK-NEXT:    vmovlb.u8 q2, q2
-; CHECK-NEXT:    vpsel q2, q2, q3
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[9]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[10]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[11]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[12]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[13]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[14]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[15]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r1
-; CHECK-NEXT:    vmovlb.u8 q0, q1
-; CHECK-NEXT:    vpt.i16 ne, q3, zr
-; CHECK-NEXT:    vaddt.i16 q2, q2, q0
-; CHECK-NEXT:    vaddva.u16 r0, q2
+; CHECK-NEXT:    vpt.i8 eq, q1, zr
+; CHECK-NEXT:    vaddvat.u8 r0, q0
 ; CHECK-NEXT:    uxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -2260,82 +2038,8 @@ entry:
 define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_acc_sext(<16 x i8> %x, <16 x i8> %b, i16 %a) {
 ; CHECK-LABEL: add_v16i8_v16i16_acc_sext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vcmp.i8 eq, q1, zr
-; CHECK-NEXT:    vmov.i8 q1, #0x0
-; CHECK-NEXT:    vmov.i8 q2, #0xff
-; CHECK-NEXT:    vmov.i32 q3, #0x0
-; CHECK-NEXT:    vpsel q1, q2, q1
-; CHECK-NEXT:    vmov.u8 r1, q1[0]
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[1]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[2]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[3]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[4]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[5]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[6]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[7]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[0]
-; CHECK-NEXT:    vcmp.i16 ne, q2, zr
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[1]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[2]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[3]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[4]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[5]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[6]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[7]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[8]
-; CHECK-NEXT:    vmovlb.s8 q2, q2
-; CHECK-NEXT:    vpsel q2, q2, q3
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[9]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[10]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[11]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[12]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[13]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[14]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[15]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r1
-; CHECK-NEXT:    vmovlb.s8 q0, q1
-; CHECK-NEXT:    vpt.i16 ne, q3, zr
-; CHECK-NEXT:    vaddt.i16 q2, q2, q0
-; CHECK-NEXT:    vaddva.u16 r0, q2
+; CHECK-NEXT:    vpt.i8 eq, q1, zr
+; CHECK-NEXT:    vaddvat.s8 r0, q0
 ; CHECK-NEXT:    sxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
diff --git a/llvm/test/CodeGen/Thumb2/mve-vecreduce-loops.ll b/llvm/test/CodeGen/Thumb2/mve-vecreduce-loops.ll
index 094f81e..539f760 100644
--- a/llvm/test/CodeGen/Thumb2/mve-vecreduce-loops.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vecreduce-loops.ll
@@ -2127,109 +2127,21 @@ define signext i16 @add16i16(i8* noalias nocapture readonly %x, i32 %n) {
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    .save {r7, lr}
 ; CHECK-NEXT:    push {r7, lr}
-; CHECK-NEXT:    .vsave {d8, d9, d10, d11, d12, d13}
-; CHECK-NEXT:    vpush {d8, d9, d10, d11, d12, d13}
-; CHECK-NEXT:    cmp r1, #0
-; CHECK-NEXT:    beq.w .LBB25_3
+; CHECK-NEXT:    cbz r1, .LBB25_4
 ; CHECK-NEXT:  @ %bb.1: @ %vector.ph
-; CHECK-NEXT:    add.w r2, r1, #15
-; CHECK-NEXT:    movs r3, #1
-; CHECK-NEXT:    bic r2, r2, #15
-; CHECK-NEXT:    vmov.i8 q0, #0x0
-; CHECK-NEXT:    subs r2, #16
-; CHECK-NEXT:    vmov.i8 q1, #0xff
-; CHECK-NEXT:    vmov.i32 q2, #0x0
-; CHECK-NEXT:    add.w lr, r3, r2, lsr #4
 ; CHECK-NEXT:    movs r2, #0
-; CHECK-NEXT:    dls lr, lr
+; CHECK-NEXT:    dlstp.8 lr, r1
 ; CHECK-NEXT:  .LBB25_2: @ %vector.body
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    vmov q4, q0
-; CHECK-NEXT:    vctp.8 r1
-; CHECK-NEXT:    vpstt
-; CHECK-NEXT:    vmovt q4, q1
-; CHECK-NEXT:    vldrbt.u8 q3, [r0], #16
-; CHECK-NEXT:    vmov.u8 r3, q4[0]
-; CHECK-NEXT:    vmov.16 q5[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[1]
-; CHECK-NEXT:    vmov.16 q5[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[2]
-; CHECK-NEXT:    vmov.16 q5[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[3]
-; CHECK-NEXT:    vmov.16 q5[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[4]
-; CHECK-NEXT:    vmov.16 q5[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[5]
-; CHECK-NEXT:    vmov.16 q5[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[6]
-; CHECK-NEXT:    vmov.16 q5[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[7]
-; CHECK-NEXT:    vmov.16 q5[7], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[0]
-; CHECK-NEXT:    vcmp.i16 ne, q5, zr
-; CHECK-NEXT:    vmov.16 q5[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[1]
-; CHECK-NEXT:    subs r1, #16
-; CHECK-NEXT:    vmov.16 q5[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[2]
-; CHECK-NEXT:    vmov.16 q5[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[3]
-; CHECK-NEXT:    vmov.16 q5[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[4]
-; CHECK-NEXT:    vmov.16 q5[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[5]
-; CHECK-NEXT:    vmov.16 q5[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[6]
-; CHECK-NEXT:    vmov.16 q5[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[7]
-; CHECK-NEXT:    vmov.16 q5[7], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[8]
-; CHECK-NEXT:    vmovlb.u8 q6, q5
-; CHECK-NEXT:    vmov q5, q2
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vmovt q5, q6
-; CHECK-NEXT:    vmov.16 q6[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[9]
-; CHECK-NEXT:    vmov.16 q6[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[10]
-; CHECK-NEXT:    vmov.16 q6[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[11]
-; CHECK-NEXT:    vmov.16 q6[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[12]
-; CHECK-NEXT:    vmov.16 q6[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[13]
-; CHECK-NEXT:    vmov.16 q6[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[14]
-; CHECK-NEXT:    vmov.16 q6[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[15]
-; CHECK-NEXT:    vmov.16 q6[7], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[8]
-; CHECK-NEXT:    vmov.16 q4[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[9]
-; CHECK-NEXT:    vmov.16 q4[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[10]
-; CHECK-NEXT:    vmov.16 q4[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[11]
-; CHECK-NEXT:    vmov.16 q4[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[12]
-; CHECK-NEXT:    vmov.16 q4[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[13]
-; CHECK-NEXT:    vmov.16 q4[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[14]
-; CHECK-NEXT:    vmov.16 q4[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[15]
-; CHECK-NEXT:    vmov.16 q4[7], r3
-; CHECK-NEXT:    vmovlb.u8 q3, q4
-; CHECK-NEXT:    vpt.i16 ne, q6, zr
-; CHECK-NEXT:    vaddt.i16 q5, q5, q3
-; CHECK-NEXT:    vaddva.u16 r2, q5
-; CHECK-NEXT:    le lr, .LBB25_2
-; CHECK-NEXT:    b .LBB25_4
-; CHECK-NEXT:  .LBB25_3:
+; CHECK-NEXT:    vldrb.u8 q0, [r0], #16
+; CHECK-NEXT:    vaddva.u8 r2, q0
+; CHECK-NEXT:    letp lr, .LBB25_2
+; CHECK-NEXT:  @ %bb.3: @ %for.cond.cleanup
+; CHECK-NEXT:    sxth r0, r2
+; CHECK-NEXT:    pop {r7, pc}
+; CHECK-NEXT:  .LBB25_4:
 ; CHECK-NEXT:    movs r2, #0
-; CHECK-NEXT:  .LBB25_4: @ %for.cond.cleanup
 ; CHECK-NEXT:    sxth r0, r2
-; CHECK-NEXT:    vpop {d8, d9, d10, d11, d12, d13}
 ; CHECK-NEXT:    pop {r7, pc}
 entry:
   %cmp8.not = icmp eq i32 %n, 0
@@ -2266,148 +2178,22 @@ define signext i16 @mla16i16(i8* noalias nocapture readonly %x, i8* noalias noca
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    .save {r7, lr}
 ; CHECK-NEXT:    push {r7, lr}
-; CHECK-NEXT:    .vsave {d8, d9, d10, d11, d12, d13, d14, d15}
-; CHECK-NEXT:    vpush {d8, d9, d10, d11, d12, d13, d14, d15}
-; CHECK-NEXT:    .pad #24
-; CHECK-NEXT:    sub sp, #24
-; CHECK-NEXT:    cmp r2, #0
-; CHECK-NEXT:    beq.w .LBB26_3
+; CHECK-NEXT:    cbz r2, .LBB26_4
 ; CHECK-NEXT:  @ %bb.1: @ %vector.ph
-; CHECK-NEXT:    add.w r3, r2, #15
-; CHECK-NEXT:    vmov.i8 q0, #0x0
-; CHECK-NEXT:    bic r3, r3, #15
-; CHECK-NEXT:    vmov.i8 q1, #0xff
-; CHECK-NEXT:    sub.w r12, r3, #16
-; CHECK-NEXT:    movs r3, #1
-; CHECK-NEXT:    vmov.i32 q2, #0x0
-; CHECK-NEXT:    add.w lr, r3, r12, lsr #4
 ; CHECK-NEXT:    mov.w r12, #0
-; CHECK-NEXT:    dls lr, lr
-; CHECK-NEXT:    vstrw.32 q1, [sp] @ 16-byte Spill
+; CHECK-NEXT:    dlstp.8 lr, r2
 ; CHECK-NEXT:  .LBB26_2: @ %vector.body
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    vmov q3, q0
-; CHECK-NEXT:    vldrw.u32 q1, [sp] @ 16-byte Reload
-; CHECK-NEXT:    vctp.8 r2
-; CHECK-NEXT:    vpsttt
-; CHECK-NEXT:    vmovt q3, q1
-; CHECK-NEXT:    vldrbt.u8 q4, [r1], #16
-; CHECK-NEXT:    vldrbt.u8 q5, [r0], #16
-; CHECK-NEXT:    vmov.u8 r3, q3[0]
-; CHECK-NEXT:    vmov q6, q2
-; CHECK-NEXT:    vmov.16 q1[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[1]
-; CHECK-NEXT:    vmov.16 q1[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[2]
-; CHECK-NEXT:    vmov.16 q1[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[3]
-; CHECK-NEXT:    vmov.16 q1[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[4]
-; CHECK-NEXT:    vmov.16 q1[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[5]
-; CHECK-NEXT:    vmov.16 q1[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[6]
-; CHECK-NEXT:    vmov.16 q1[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[7]
-; CHECK-NEXT:    vmov.16 q1[7], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[0]
-; CHECK-NEXT:    vcmp.i16 ne, q1, zr
-; CHECK-NEXT:    vmov.16 q1[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[1]
-; CHECK-NEXT:    subs r2, #16
-; CHECK-NEXT:    vmov.16 q1[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[2]
-; CHECK-NEXT:    vmov.16 q1[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[3]
-; CHECK-NEXT:    vmov.16 q1[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[4]
-; CHECK-NEXT:    vmov.16 q1[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[5]
-; CHECK-NEXT:    vmov.16 q1[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[6]
-; CHECK-NEXT:    vmov.16 q1[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[7]
-; CHECK-NEXT:    vmov.16 q1[7], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[0]
-; CHECK-NEXT:    vmovlb.u8 q7, q1
-; CHECK-NEXT:    vmov.16 q1[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[1]
-; CHECK-NEXT:    vmov.16 q1[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[2]
-; CHECK-NEXT:    vmov.16 q1[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[3]
-; CHECK-NEXT:    vmov.16 q1[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[4]
-; CHECK-NEXT:    vmov.16 q1[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[5]
-; CHECK-NEXT:    vmov.16 q1[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[6]
-; CHECK-NEXT:    vmov.16 q1[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[7]
-; CHECK-NEXT:    vmov.16 q1[7], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[8]
-; CHECK-NEXT:    vmovlb.u8 q1, q1
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vmult.i16 q6, q1, q7
-; CHECK-NEXT:    vmov.16 q7[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[9]
-; CHECK-NEXT:    vmov.16 q7[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[10]
-; CHECK-NEXT:    vmov.16 q7[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[11]
-; CHECK-NEXT:    vmov.16 q7[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[12]
-; CHECK-NEXT:    vmov.16 q7[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[13]
-; CHECK-NEXT:    vmov.16 q7[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[14]
-; CHECK-NEXT:    vmov.16 q7[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[8]
-; CHECK-NEXT:    vmov.16 q1[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[9]
-; CHECK-NEXT:    vmov.16 q1[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[10]
-; CHECK-NEXT:    vmov.16 q1[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[11]
-; CHECK-NEXT:    vmov.16 q1[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[12]
-; CHECK-NEXT:    vmov.16 q1[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[13]
-; CHECK-NEXT:    vmov.16 q1[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[14]
-; CHECK-NEXT:    vmov.16 q1[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q5[15]
-; CHECK-NEXT:    vmov.16 q1[7], r3
-; CHECK-NEXT:    vmov.u8 r3, q4[15]
-; CHECK-NEXT:    vmov.16 q7[7], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[8]
-; CHECK-NEXT:    vmullb.u8 q4, q7, q1
-; CHECK-NEXT:    vmov.16 q1[0], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[9]
-; CHECK-NEXT:    vmov.16 q1[1], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[10]
-; CHECK-NEXT:    vmov.16 q1[2], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[11]
-; CHECK-NEXT:    vmov.16 q1[3], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[12]
-; CHECK-NEXT:    vmov.16 q1[4], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[13]
-; CHECK-NEXT:    vmov.16 q1[5], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[14]
-; CHECK-NEXT:    vmov.16 q1[6], r3
-; CHECK-NEXT:    vmov.u8 r3, q3[15]
-; CHECK-NEXT:    vmov.16 q1[7], r3
-; CHECK-NEXT:    vpt.i16 ne, q1, zr
-; CHECK-NEXT:    vaddt.i16 q6, q6, q4
-; CHECK-NEXT:    vaddva.u16 r12, q6
-; CHECK-NEXT:    le lr, .LBB26_2
-; CHECK-NEXT:    b .LBB26_4
-; CHECK-NEXT:  .LBB26_3:
+; CHECK-NEXT:    vldrb.u8 q0, [r0], #16
+; CHECK-NEXT:    vldrb.u8 q1, [r1], #16
+; CHECK-NEXT:    vmlava.u8 r12, q1, q0
+; CHECK-NEXT:    letp lr, .LBB26_2
+; CHECK-NEXT:  @ %bb.3: @ %for.cond.cleanup
+; CHECK-NEXT:    sxth.w r0, r12
+; CHECK-NEXT:    pop {r7, pc}
+; CHECK-NEXT:  .LBB26_4:
 ; CHECK-NEXT:    mov.w r12, #0
-; CHECK-NEXT:  .LBB26_4: @ %for.cond.cleanup
 ; CHECK-NEXT:    sxth.w r0, r12
-; CHECK-NEXT:    add sp, #24
-; CHECK-NEXT:    vpop {d8, d9, d10, d11, d12, d13, d14, d15}
 ; CHECK-NEXT:    pop {r7, pc}
 entry:
   %cmp13.not = icmp eq i32 %n, 0
diff --git a/llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll b/llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll
index 67a0075..93e3b16 100644
--- a/llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll
@@ -275,74 +275,7 @@ entry:
 define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_zext(<16 x i8> %x, <16 x i8> %y) {
 ; CHECK-LABEL: add_v16i8_v16i16_zext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[9]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[11]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[13]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[15]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[8]
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[9]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[10]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[11]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[12]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[13]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[14]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[15]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vmullb.u8 q2, q3, q2
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[1]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[3]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[5]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[7]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vmov.16 q1[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[1]
-; CHECK-NEXT:    vmov.16 q1[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[2]
-; CHECK-NEXT:    vmov.16 q1[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[3]
-; CHECK-NEXT:    vmov.16 q1[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[4]
-; CHECK-NEXT:    vmov.16 q1[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[5]
-; CHECK-NEXT:    vmov.16 q1[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[6]
-; CHECK-NEXT:    vmov.16 q1[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[7]
-; CHECK-NEXT:    vmov.16 q1[7], r0
-; CHECK-NEXT:    vmullb.u8 q0, q1, q3
-; CHECK-NEXT:    vadd.i16 q0, q0, q2
-; CHECK-NEXT:    vaddv.u16 r0, q0
+; CHECK-NEXT:    vmlav.u8 r0, q0, q1
 ; CHECK-NEXT:    uxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -356,74 +289,7 @@ entry:
 define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_sext(<16 x i8> %x, <16 x i8> %y) {
 ; CHECK-LABEL: add_v16i8_v16i16_sext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[9]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[11]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[13]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[15]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[8]
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[9]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[10]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[11]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[12]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[13]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[14]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[15]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vmullb.s8 q2, q3, q2
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[1]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[3]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[5]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[7]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vmov.16 q1[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[1]
-; CHECK-NEXT:    vmov.16 q1[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[2]
-; CHECK-NEXT:    vmov.16 q1[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[3]
-; CHECK-NEXT:    vmov.16 q1[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[4]
-; CHECK-NEXT:    vmov.16 q1[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[5]
-; CHECK-NEXT:    vmov.16 q1[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[6]
-; CHECK-NEXT:    vmov.16 q1[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[7]
-; CHECK-NEXT:    vmov.16 q1[7], r0
-; CHECK-NEXT:    vmullb.s8 q0, q1, q3
-; CHECK-NEXT:    vadd.i16 q0, q0, q2
-; CHECK-NEXT:    vaddv.u16 r0, q0
+; CHECK-NEXT:    vmlav.s8 r0, q0, q1
 ; CHECK-NEXT:    sxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -1243,74 +1109,7 @@ entry:
 define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_acc_zext(<16 x i8> %x, <16 x i8> %y, i16 %a) {
 ; CHECK-LABEL: add_v16i8_v16i16_acc_zext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r1, q1[8]
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[9]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[10]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[11]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[12]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[13]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[14]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[15]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[8]
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[9]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[10]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[11]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[12]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[13]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[14]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[15]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[0]
-; CHECK-NEXT:    vmullb.u8 q2, q3, q2
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[1]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[2]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[3]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[4]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[5]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[6]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[7]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[0]
-; CHECK-NEXT:    vmov.16 q1[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[1]
-; CHECK-NEXT:    vmov.16 q1[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[2]
-; CHECK-NEXT:    vmov.16 q1[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[3]
-; CHECK-NEXT:    vmov.16 q1[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[4]
-; CHECK-NEXT:    vmov.16 q1[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[5]
-; CHECK-NEXT:    vmov.16 q1[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[6]
-; CHECK-NEXT:    vmov.16 q1[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[7]
-; CHECK-NEXT:    vmov.16 q1[7], r1
-; CHECK-NEXT:    vmullb.u8 q0, q1, q3
-; CHECK-NEXT:    vadd.i16 q0, q0, q2
-; CHECK-NEXT:    vaddva.u16 r0, q0
+; CHECK-NEXT:    vmlava.u8 r0, q0, q1
 ; CHECK-NEXT:    uxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
@@ -1325,74 +1124,7 @@ entry:
 define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_acc_sext(<16 x i8> %x, <16 x i8> %y, i16 %a) {
 ; CHECK-LABEL: add_v16i8_v16i16_acc_sext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    vmov.u8 r1, q1[8]
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[9]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[10]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[11]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[12]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[13]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[14]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[15]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[8]
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[9]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[10]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[11]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[12]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[13]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[14]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[15]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[0]
-; CHECK-NEXT:    vmullb.s8 q2, q3, q2
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[1]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[2]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[3]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[4]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[5]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[6]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[7]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[0]
-; CHECK-NEXT:    vmov.16 q1[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[1]
-; CHECK-NEXT:    vmov.16 q1[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[2]
-; CHECK-NEXT:    vmov.16 q1[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[3]
-; CHECK-NEXT:    vmov.16 q1[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[4]
-; CHECK-NEXT:    vmov.16 q1[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[5]
-; CHECK-NEXT:    vmov.16 q1[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[6]
-; CHECK-NEXT:    vmov.16 q1[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[7]
-; CHECK-NEXT:    vmov.16 q1[7], r1
-; CHECK-NEXT:    vmullb.s8 q0, q1, q3
-; CHECK-NEXT:    vadd.i16 q0, q0, q2
-; CHECK-NEXT:    vaddva.u16 r0, q0
+; CHECK-NEXT:    vmlava.s8 r0, q0, q1
 ; CHECK-NEXT:    sxth r0, r0
 ; CHECK-NEXT:    bx lr
 entry:
diff --git a/llvm/test/CodeGen/Thumb2/mve-vecreduce-mlapred.ll b/llvm/test/CodeGen/Thumb2/mve-vecreduce-mlapred.ll
index 6510aff..f30856d 100644
--- a/llvm/test/CodeGen/Thumb2/mve-vecreduce-mlapred.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vecreduce-mlapred.ll
@@ -416,120 +416,9 @@ entry:
 define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_zext(<16 x i8> %x, <16 x i8> %y, <16 x i8> %b) {
 ; CHECK-LABEL: add_v16i8_v16i16_zext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    .vsave {d8, d9, d10, d11}
-; CHECK-NEXT:    vpush {d8, d9, d10, d11}
-; CHECK-NEXT:    vcmp.i8 eq, q2, zr
-; CHECK-NEXT:    vmov.i8 q2, #0x0
-; CHECK-NEXT:    vmov.i8 q3, #0xff
-; CHECK-NEXT:    vpsel q2, q3, q2
-; CHECK-NEXT:    vmov.u8 r0, q2[0]
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[1]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[2]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[3]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[4]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[5]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[6]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[7]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vcmp.i16 ne, q3, zr
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[1]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[3]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[5]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[7]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vmovlb.u8 q4, q3
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[1]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[2]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[3]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[4]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[5]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[6]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[7]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[8]
-; CHECK-NEXT:    vmovlb.u8 q5, q3
-; CHECK-NEXT:    vmov.i32 q3, #0x0
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vmult.i16 q3, q5, q4
-; CHECK-NEXT:    vmov.16 q4[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[9]
-; CHECK-NEXT:    vmov.16 q4[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[10]
-; CHECK-NEXT:    vmov.16 q4[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[11]
-; CHECK-NEXT:    vmov.16 q4[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[12]
-; CHECK-NEXT:    vmov.16 q4[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[13]
-; CHECK-NEXT:    vmov.16 q4[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[14]
-; CHECK-NEXT:    vmov.16 q4[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[15]
-; CHECK-NEXT:    vmov.16 q4[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[9]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[11]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[13]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[15]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r0
-; CHECK-NEXT:    vmullb.u8 q0, q1, q2
-; CHECK-NEXT:    vpt.i16 ne, q4, zr
-; CHECK-NEXT:    vaddt.i16 q3, q3, q0
-; CHECK-NEXT:    vaddv.u16 r0, q3
+; CHECK-NEXT:    vpt.i8 eq, q2, zr
+; CHECK-NEXT:    vmlavt.u8 r0, q0, q1
 ; CHECK-NEXT:    uxth r0, r0
-; CHECK-NEXT:    vpop {d8, d9, d10, d11}
 ; CHECK-NEXT:    bx lr
 entry:
   %c = icmp eq <16 x i8> %b, zeroinitializer
@@ -544,120 +433,9 @@ entry:
 define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_sext(<16 x i8> %x, <16 x i8> %y, <16 x i8> %b) {
 ; CHECK-LABEL: add_v16i8_v16i16_sext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    .vsave {d8, d9, d10, d11}
-; CHECK-NEXT:    vpush {d8, d9, d10, d11}
-; CHECK-NEXT:    vcmp.i8 eq, q2, zr
-; CHECK-NEXT:    vmov.i8 q2, #0x0
-; CHECK-NEXT:    vmov.i8 q3, #0xff
-; CHECK-NEXT:    vpsel q2, q3, q2
-; CHECK-NEXT:    vmov.u8 r0, q2[0]
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[1]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[2]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[3]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[4]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[5]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[6]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[7]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[0]
-; CHECK-NEXT:    vcmp.i16 ne, q3, zr
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[1]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[2]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[3]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[4]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[5]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[6]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[7]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[0]
-; CHECK-NEXT:    vmovlb.s8 q4, q3
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[1]
-; CHECK-NEXT:    vmov.16 q3[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[2]
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[3]
-; CHECK-NEXT:    vmov.16 q3[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[4]
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[5]
-; CHECK-NEXT:    vmov.16 q3[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[6]
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[7]
-; CHECK-NEXT:    vmov.16 q3[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[8]
-; CHECK-NEXT:    vmovlb.s8 q5, q3
-; CHECK-NEXT:    vmov.i32 q3, #0x0
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vmult.i16 q3, q5, q4
-; CHECK-NEXT:    vmov.16 q4[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[9]
-; CHECK-NEXT:    vmov.16 q4[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[10]
-; CHECK-NEXT:    vmov.16 q4[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[11]
-; CHECK-NEXT:    vmov.16 q4[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[12]
-; CHECK-NEXT:    vmov.16 q4[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[13]
-; CHECK-NEXT:    vmov.16 q4[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[14]
-; CHECK-NEXT:    vmov.16 q4[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q2[15]
-; CHECK-NEXT:    vmov.16 q4[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[8]
-; CHECK-NEXT:    vmov.16 q2[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[9]
-; CHECK-NEXT:    vmov.16 q2[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[10]
-; CHECK-NEXT:    vmov.16 q2[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[11]
-; CHECK-NEXT:    vmov.16 q2[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[12]
-; CHECK-NEXT:    vmov.16 q2[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[13]
-; CHECK-NEXT:    vmov.16 q2[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[14]
-; CHECK-NEXT:    vmov.16 q2[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q1[15]
-; CHECK-NEXT:    vmov.16 q2[7], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r0
-; CHECK-NEXT:    vmov.u8 r0, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r0
-; CHECK-NEXT:    vmullb.s8 q0, q1, q2
-; CHECK-NEXT:    vpt.i16 ne, q4, zr
-; CHECK-NEXT:    vaddt.i16 q3, q3, q0
-; CHECK-NEXT:    vaddv.u16 r0, q3
+; CHECK-NEXT:    vpt.i8 eq, q2, zr
+; CHECK-NEXT:    vmlavt.s8 r0, q0, q1
 ; CHECK-NEXT:    sxth r0, r0
-; CHECK-NEXT:    vpop {d8, d9, d10, d11}
 ; CHECK-NEXT:    bx lr
 entry:
   %c = icmp eq <16 x i8> %b, zeroinitializer
@@ -2084,120 +1862,9 @@ entry:
 define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_acc_zext(<16 x i8> %x, <16 x i8> %y, <16 x i8> %b, i16 %a) {
 ; CHECK-LABEL: add_v16i8_v16i16_acc_zext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    .vsave {d8, d9, d10, d11}
-; CHECK-NEXT:    vpush {d8, d9, d10, d11}
-; CHECK-NEXT:    vcmp.i8 eq, q2, zr
-; CHECK-NEXT:    vmov.i8 q2, #0x0
-; CHECK-NEXT:    vmov.i8 q3, #0xff
-; CHECK-NEXT:    vpsel q2, q3, q2
-; CHECK-NEXT:    vmov.u8 r1, q2[0]
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[1]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[2]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[3]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[4]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[5]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[6]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[7]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[0]
-; CHECK-NEXT:    vcmp.i16 ne, q3, zr
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[1]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[2]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[3]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[4]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[5]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[6]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[7]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[0]
-; CHECK-NEXT:    vmovlb.u8 q4, q3
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[1]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[2]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[3]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[4]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[5]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[6]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[7]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[8]
-; CHECK-NEXT:    vmovlb.u8 q5, q3
-; CHECK-NEXT:    vmov.i32 q3, #0x0
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vmult.i16 q3, q5, q4
-; CHECK-NEXT:    vmov.16 q4[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[9]
-; CHECK-NEXT:    vmov.16 q4[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[10]
-; CHECK-NEXT:    vmov.16 q4[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[11]
-; CHECK-NEXT:    vmov.16 q4[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[12]
-; CHECK-NEXT:    vmov.16 q4[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[13]
-; CHECK-NEXT:    vmov.16 q4[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[14]
-; CHECK-NEXT:    vmov.16 q4[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[15]
-; CHECK-NEXT:    vmov.16 q4[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[8]
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[9]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[10]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[11]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[12]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[13]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[14]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[15]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r1
-; CHECK-NEXT:    vmullb.u8 q0, q1, q2
-; CHECK-NEXT:    vpt.i16 ne, q4, zr
-; CHECK-NEXT:    vaddt.i16 q3, q3, q0
-; CHECK-NEXT:    vaddva.u16 r0, q3
+; CHECK-NEXT:    vpt.i8 eq, q2, zr
+; CHECK-NEXT:    vmlavat.u8 r0, q0, q1
 ; CHECK-NEXT:    uxth r0, r0
-; CHECK-NEXT:    vpop {d8, d9, d10, d11}
 ; CHECK-NEXT:    bx lr
 entry:
   %c = icmp eq <16 x i8> %b, zeroinitializer
@@ -2213,120 +1880,9 @@ entry:
 define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_acc_sext(<16 x i8> %x, <16 x i8> %y, <16 x i8> %b, i16 %a) {
 ; CHECK-LABEL: add_v16i8_v16i16_acc_sext:
 ; CHECK:       @ %bb.0: @ %entry
-; CHECK-NEXT:    .vsave {d8, d9, d10, d11}
-; CHECK-NEXT:    vpush {d8, d9, d10, d11}
-; CHECK-NEXT:    vcmp.i8 eq, q2, zr
-; CHECK-NEXT:    vmov.i8 q2, #0x0
-; CHECK-NEXT:    vmov.i8 q3, #0xff
-; CHECK-NEXT:    vpsel q2, q3, q2
-; CHECK-NEXT:    vmov.u8 r1, q2[0]
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[1]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[2]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[3]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[4]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[5]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[6]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[7]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[0]
-; CHECK-NEXT:    vcmp.i16 ne, q3, zr
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[1]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[2]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[3]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[4]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[5]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[6]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[7]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[0]
-; CHECK-NEXT:    vmovlb.s8 q4, q3
-; CHECK-NEXT:    vmov.16 q3[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[1]
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[2]
-; CHECK-NEXT:    vmov.16 q3[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[3]
-; CHECK-NEXT:    vmov.16 q3[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[4]
-; CHECK-NEXT:    vmov.16 q3[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[5]
-; CHECK-NEXT:    vmov.16 q3[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[6]
-; CHECK-NEXT:    vmov.16 q3[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[7]
-; CHECK-NEXT:    vmov.16 q3[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[8]
-; CHECK-NEXT:    vmovlb.s8 q5, q3
-; CHECK-NEXT:    vmov.i32 q3, #0x0
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vmult.i16 q3, q5, q4
-; CHECK-NEXT:    vmov.16 q4[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[9]
-; CHECK-NEXT:    vmov.16 q4[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[10]
-; CHECK-NEXT:    vmov.16 q4[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[11]
-; CHECK-NEXT:    vmov.16 q4[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[12]
-; CHECK-NEXT:    vmov.16 q4[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[13]
-; CHECK-NEXT:    vmov.16 q4[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[14]
-; CHECK-NEXT:    vmov.16 q4[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q2[15]
-; CHECK-NEXT:    vmov.16 q4[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[8]
-; CHECK-NEXT:    vmov.16 q2[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[9]
-; CHECK-NEXT:    vmov.16 q2[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[10]
-; CHECK-NEXT:    vmov.16 q2[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[11]
-; CHECK-NEXT:    vmov.16 q2[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[12]
-; CHECK-NEXT:    vmov.16 q2[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[13]
-; CHECK-NEXT:    vmov.16 q2[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[14]
-; CHECK-NEXT:    vmov.16 q2[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q1[15]
-; CHECK-NEXT:    vmov.16 q2[7], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[8]
-; CHECK-NEXT:    vmov.16 q1[0], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[9]
-; CHECK-NEXT:    vmov.16 q1[1], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[10]
-; CHECK-NEXT:    vmov.16 q1[2], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[11]
-; CHECK-NEXT:    vmov.16 q1[3], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[12]
-; CHECK-NEXT:    vmov.16 q1[4], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[13]
-; CHECK-NEXT:    vmov.16 q1[5], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[14]
-; CHECK-NEXT:    vmov.16 q1[6], r1
-; CHECK-NEXT:    vmov.u8 r1, q0[15]
-; CHECK-NEXT:    vmov.16 q1[7], r1
-; CHECK-NEXT:    vmullb.s8 q0, q1, q2
-; CHECK-NEXT:    vpt.i16 ne, q4, zr
-; CHECK-NEXT:    vaddt.i16 q3, q3, q0
-; CHECK-NEXT:    vaddva.u16 r0, q3
+; CHECK-NEXT:    vpt.i8 eq, q2, zr
+; CHECK-NEXT:    vmlavat.s8 r0, q0, q1
 ; CHECK-NEXT:    sxth r0, r0
-; CHECK-NEXT:    vpop {d8, d9, d10, d11}
 ; CHECK-NEXT:    bx lr
 entry:
   %c = icmp eq <16 x i8> %b, zeroinitializer
-- 
2.7.4