[llvm] [LV]: Ensure fairness when selecting epilogue VF. (PR #155547)

Hassnaa Hamdi via llvm-commits llvm-commits at lists.llvm.org
Tue Sep 9 08:54:35 PDT 2025


https://github.com/hassnaaHamdi updated https://github.com/llvm/llvm-project/pull/155547

>From c168005ec1d87bc19f8e41574a866f20ca7b3ef1 Mon Sep 17 00:00:00 2001
From: Hassnaa Hamdi <hassnaa.hamdi at arm.com>
Date: Tue, 26 Aug 2025 08:50:00 +0000
Subject: [PATCH 1/3] [LV]: Ensure fairness when selecting epilogue VF.

Consider IC when deciding if epilogue profitable for scalable
vectors, same as fixed-width vectors.
---
 .../Transforms/Vectorize/LoopVectorize.cpp    |   8 +-
 .../AArch64/induction-costs-sve.ll            |  87 ++++++---
 .../AArch64/interleaving-reduction.ll         | 179 ++++++++++--------
 .../AArch64/reduction-recurrence-costs-sve.ll |  66 +++++--
 4 files changed, 218 insertions(+), 122 deletions(-)

diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 88e477e2dc64a..f4226c7f8dd86 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -4310,16 +4310,10 @@ bool LoopVectorizationCostModel::isEpilogueVectorizationProfitable(
   if (TTI.getMaxInterleaveFactor(VF) <= 1)
     return false;
 
-  // TODO: PR #108190 introduced a discrepancy between fixed-width and scalable
-  // VFs when deciding profitability.
-  // See related "TODO: extend to support scalable VFs." in
-  // selectEpilogueVectorizationFactor.
-  unsigned Multiplier = VF.isFixed() ? IC : 1;
   unsigned MinVFThreshold = EpilogueVectorizationMinVF.getNumOccurrences() > 0
                                 ? EpilogueVectorizationMinVF
                                 : TTI.getEpilogueVectorizationMinVF();
-  return estimateElementCount(VF * Multiplier, VScaleForTuning) >=
-         MinVFThreshold;
+  return estimateElementCount(VF * IC, VScaleForTuning) >= MinVFThreshold;
 }
 
 VectorizationFactor LoopVectorizationPlanner::selectEpilogueVectorizationFactor(
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll b/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
index f094b9a72d85b..02b1fc34f7fc6 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
@@ -8,24 +8,29 @@ target triple = "arm64-apple-macosx14.0.0"
 define void @iv_casts(ptr %dst, ptr %src, i32 %x, i64 %N) #0 {
 ; DEFAULT-LABEL: define void @iv_casts(
 ; DEFAULT-SAME: ptr [[DST:%.*]], ptr [[SRC:%.*]], i32 [[X:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
-; DEFAULT-NEXT:  [[ENTRY:.*]]:
+; DEFAULT-NEXT:  [[ITER_CHECK:.*]]:
 ; DEFAULT-NEXT:    [[SRC2:%.*]] = ptrtoint ptr [[SRC]] to i64
 ; DEFAULT-NEXT:    [[DST1:%.*]] = ptrtoint ptr [[DST]] to i64
 ; DEFAULT-NEXT:    [[TMP0:%.*]] = add i64 [[N]], 1
 ; DEFAULT-NEXT:    [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
-; DEFAULT-NEXT:    [[TMP2:%.*]] = shl nuw i64 [[TMP1]], 4
+; DEFAULT-NEXT:    [[TMP2:%.*]] = shl nuw i64 [[TMP1]], 2
 ; DEFAULT-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], [[TMP2]]
-; DEFAULT-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
+; DEFAULT-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
 ; DEFAULT:       [[VECTOR_MEMCHECK]]:
 ; DEFAULT-NEXT:    [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
 ; DEFAULT-NEXT:    [[TMP4:%.*]] = mul nuw i64 [[TMP3]], 8
 ; DEFAULT-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 2
 ; DEFAULT-NEXT:    [[TMP6:%.*]] = sub i64 [[DST1]], [[SRC2]]
 ; DEFAULT-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP6]], [[TMP5]]
-; DEFAULT-NEXT:    br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
-; DEFAULT:       [[VECTOR_PH]]:
+; DEFAULT-NEXT:    br i1 [[DIFF_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VECTOR_MAIN_LOOP_ITER_CHECK:.*]]
+; DEFAULT:       [[VECTOR_MAIN_LOOP_ITER_CHECK]]:
 ; DEFAULT-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
-; DEFAULT-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP9]], 16
+; DEFAULT-NEXT:    [[TMP8:%.*]] = shl nuw i64 [[TMP9]], 4
+; DEFAULT-NEXT:    [[MIN_ITERS_CHECK3:%.*]] = icmp ult i64 [[TMP0]], [[TMP8]]
+; DEFAULT-NEXT:    br i1 [[MIN_ITERS_CHECK3]], label %[[VEC_EPILOG_PH:.*]], label %[[VECTOR_PH:.*]]
+; DEFAULT:       [[VECTOR_PH]]:
+; DEFAULT-NEXT:    [[TMP11:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP11]], 16
 ; DEFAULT-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], [[TMP10]]
 ; DEFAULT-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
 ; DEFAULT-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 8 x i32> poison, i32 [[X]], i64 0
@@ -61,25 +66,59 @@ define void @iv_casts(ptr %dst, ptr %src, i32 %x, i64 %N) #0 {
 ; DEFAULT-NEXT:    br i1 [[TMP44]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
-; DEFAULT:       [[SCALAR_PH]]:
-; DEFAULT-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_MEMCHECK]] ]
+; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[VEC_EPILOG_ITER_CHECK:.*]]
+; DEFAULT:       [[VEC_EPILOG_ITER_CHECK]]:
+; DEFAULT-NEXT:    [[N_VEC_REMAINING:%.*]] = sub i64 [[TMP0]], [[N_VEC]]
+; DEFAULT-NEXT:    [[TMP47:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP48:%.*]] = shl nuw i64 [[TMP47]], 2
+; DEFAULT-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], [[TMP48]]
+; DEFAULT-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VEC_EPILOG_PH]]
+; DEFAULT:       [[VEC_EPILOG_PH]]:
+; DEFAULT-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; DEFAULT-NEXT:    [[TMP49:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP50:%.*]] = mul nuw i64 [[TMP49]], 4
+; DEFAULT-NEXT:    [[N_MOD_VF5:%.*]] = urem i64 [[TMP0]], [[TMP50]]
+; DEFAULT-NEXT:    [[N_VEC6:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF5]]
+; DEFAULT-NEXT:    [[BROADCAST_SPLATINSERT7:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[X]], i64 0
+; DEFAULT-NEXT:    [[BROADCAST_SPLAT8:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT7]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
+; DEFAULT-NEXT:    [[TMP51:%.*]] = trunc <vscale x 4 x i32> [[BROADCAST_SPLAT8]] to <vscale x 4 x i16>
+; DEFAULT-NEXT:    br label %[[VEC_EPILOG_VECTOR_BODY:.*]]
+; DEFAULT:       [[VEC_EPILOG_VECTOR_BODY]]:
+; DEFAULT-NEXT:    [[IV:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT11:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
+; DEFAULT-NEXT:    [[GEP_SRC:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[IV]]
+; DEFAULT-NEXT:    [[WIDE_LOAD10:%.*]] = load <vscale x 4 x i8>, ptr [[GEP_SRC]], align 1
+; DEFAULT-NEXT:    [[TMP39:%.*]] = zext <vscale x 4 x i8> [[WIDE_LOAD10]] to <vscale x 4 x i16>
+; DEFAULT-NEXT:    [[TMP40:%.*]] = mul <vscale x 4 x i16> [[TMP39]], [[TMP51]]
+; DEFAULT-NEXT:    [[TMP52:%.*]] = zext <vscale x 4 x i8> [[WIDE_LOAD10]] to <vscale x 4 x i16>
+; DEFAULT-NEXT:    [[TMP53:%.*]] = or <vscale x 4 x i16> [[TMP40]], [[TMP52]]
+; DEFAULT-NEXT:    [[TMP54:%.*]] = lshr <vscale x 4 x i16> [[TMP53]], splat (i16 1)
+; DEFAULT-NEXT:    [[TMP55:%.*]] = trunc <vscale x 4 x i16> [[TMP54]] to <vscale x 4 x i8>
+; DEFAULT-NEXT:    [[TMP45:%.*]] = getelementptr i8, ptr [[DST]], i64 [[IV]]
+; DEFAULT-NEXT:    store <vscale x 4 x i8> [[TMP55]], ptr [[TMP45]], align 1
+; DEFAULT-NEXT:    [[INDEX_NEXT11]] = add nuw i64 [[IV]], [[TMP50]]
+; DEFAULT-NEXT:    [[TMP46:%.*]] = icmp eq i64 [[INDEX_NEXT11]], [[N_VEC6]]
+; DEFAULT-NEXT:    br i1 [[TMP46]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; DEFAULT:       [[VEC_EPILOG_MIDDLE_BLOCK]]:
+; DEFAULT-NEXT:    [[CMP_N12:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC6]]
+; DEFAULT-NEXT:    br i1 [[CMP_N12]], label %[[EXIT]], label %[[VEC_EPILOG_SCALAR_PH]]
+; DEFAULT:       [[VEC_EPILOG_SCALAR_PH]]:
+; DEFAULT-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC6]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MEMCHECK]] ], [ 0, %[[ITER_CHECK]] ]
 ; DEFAULT-NEXT:    br label %[[LOOP:.*]]
 ; DEFAULT:       [[LOOP]]:
-; DEFAULT-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
-; DEFAULT-NEXT:    [[GEP_SRC:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[IV]]
-; DEFAULT-NEXT:    [[L:%.*]] = load i8, ptr [[GEP_SRC]], align 1
+; DEFAULT-NEXT:    [[IV1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
+; DEFAULT-NEXT:    [[GEP_SRC1:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[IV1]]
+; DEFAULT-NEXT:    [[L:%.*]] = load i8, ptr [[GEP_SRC1]], align 1
 ; DEFAULT-NEXT:    [[L_EXT:%.*]] = zext i8 [[L]] to i32
 ; DEFAULT-NEXT:    [[MUL16_US:%.*]] = mul i32 [[L_EXT]], [[X]]
-; DEFAULT-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
+; DEFAULT-NEXT:    [[IV_NEXT]] = add i64 [[IV1]], 1
 ; DEFAULT-NEXT:    [[CONV25_US:%.*]] = zext i8 [[L]] to i32
 ; DEFAULT-NEXT:    [[ADD34_US:%.*]] = or i32 [[MUL16_US]], [[CONV25_US]]
 ; DEFAULT-NEXT:    [[SHR35_US:%.*]] = lshr i32 [[ADD34_US]], 1
 ; DEFAULT-NEXT:    [[CONV36_US:%.*]] = trunc i32 [[SHR35_US]] to i8
-; DEFAULT-NEXT:    [[GEP_DST:%.*]] = getelementptr i8, ptr [[DST]], i64 [[IV]]
+; DEFAULT-NEXT:    [[GEP_DST:%.*]] = getelementptr i8, ptr [[DST]], i64 [[IV1]]
 ; DEFAULT-NEXT:    store i8 [[CONV36_US]], ptr [[GEP_DST]], align 1
-; DEFAULT-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV]], [[N]]
-; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
+; DEFAULT-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV1]], [[N]]
+; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP4:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -215,7 +254,7 @@ define void @iv_trunc(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 1, ptr [[TMP21]], align 4
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
 ; DEFAULT-NEXT:    [[TMP22:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP22]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[TMP22]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -231,7 +270,7 @@ define void @iv_trunc(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 1, ptr [[GEP]], align 4
 ; DEFAULT-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
 ; DEFAULT-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV]], [[N]]
-; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -379,7 +418,7 @@ define void @trunc_ivs_and_store(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 [[TMP15]], ptr [[TMP24]], align 4
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
 ; DEFAULT-NEXT:    [[TMP25:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP25]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[TMP25]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -398,7 +437,7 @@ define void @trunc_ivs_and_store(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 [[IV_2]], ptr [[GEP]], align 4
 ; DEFAULT-NEXT:    [[IV_1_NEXT]] = add i64 [[IV_1]], 1
 ; DEFAULT-NEXT:    [[EXITCOND_3_NOT:%.*]] = icmp eq i64 [[IV_1]], [[N]]
-; DEFAULT-NEXT:    br i1 [[EXITCOND_3_NOT]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[EXITCOND_3_NOT]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP8:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -571,7 +610,7 @@ define void @ivs_trunc_and_ext(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 [[TMP14]], ptr [[TMP23]], align 4
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
 ; DEFAULT-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP24]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[TMP24]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -590,7 +629,7 @@ define void @ivs_trunc_and_ext(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 [[IV_2]], ptr [[GEP]], align 4
 ; DEFAULT-NEXT:    [[IV_1_NEXT]] = add i64 [[IV_1]], 1
 ; DEFAULT-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_1]], [[N]]
-; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP10:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -747,7 +786,7 @@ define void @exit_cond_zext_iv(ptr %dst, i64 %N) {
 ; DEFAULT-NEXT:    store i32 0, ptr [[TMP10]], align 8
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
 ; DEFAULT-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP11]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[TMP11]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[UMAX1]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -763,7 +802,7 @@ define void @exit_cond_zext_iv(ptr %dst, i64 %N) {
 ; DEFAULT-NEXT:    [[IV_1_NEXT]] = add i32 [[IV_1]], 1
 ; DEFAULT-NEXT:    [[IV_EXT]] = zext i32 [[IV_1_NEXT]] to i64
 ; DEFAULT-NEXT:    [[C:%.*]] = icmp ult i64 [[IV_EXT]], [[N]]
-; DEFAULT-NEXT:    br i1 [[C]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP11:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[C]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP12:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/interleaving-reduction.ll b/llvm/test/Transforms/LoopVectorize/AArch64/interleaving-reduction.ll
index 7ff4609f8ec4b..1acd1897fa1f2 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/interleaving-reduction.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/interleaving-reduction.ll
@@ -24,30 +24,30 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-4-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; INTERLEAVE-4:       vector.body:
 ; INTERLEAVE-4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP12:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI2:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI3:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI4:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
-; INTERLEAVE-4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 4
-; INTERLEAVE-4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 8
-; INTERLEAVE-4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 12
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP4]], align 1
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD5:%.*]] = load <4 x i32>, ptr [[TMP9]], align 1
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD6:%.*]] = load <4 x i32>, ptr [[TMP10]], align 1
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP11]], align 1
-; INTERLEAVE-4-NEXT:    [[TMP12]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
-; INTERLEAVE-4-NEXT:    [[TMP13]] = add <4 x i32> [[VEC_PHI2]], [[WIDE_LOAD5]]
-; INTERLEAVE-4-NEXT:    [[TMP14]] = add <4 x i32> [[VEC_PHI3]], [[WIDE_LOAD6]]
-; INTERLEAVE-4-NEXT:    [[TMP15]] = add <4 x i32> [[VEC_PHI4]], [[WIDE_LOAD7]]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI2:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI3:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI4:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
+; INTERLEAVE-4-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 4
+; INTERLEAVE-4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 8
+; INTERLEAVE-4-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 12
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 1
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD5:%.*]] = load <4 x i32>, ptr [[TMP1]], align 1
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD6:%.*]] = load <4 x i32>, ptr [[TMP2]], align 1
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP3]], align 1
+; INTERLEAVE-4-NEXT:    [[TMP4]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
+; INTERLEAVE-4-NEXT:    [[TMP5]] = add <4 x i32> [[VEC_PHI2]], [[WIDE_LOAD5]]
+; INTERLEAVE-4-NEXT:    [[TMP6]] = add <4 x i32> [[VEC_PHI3]], [[WIDE_LOAD6]]
+; INTERLEAVE-4-NEXT:    [[TMP7]] = add <4 x i32> [[VEC_PHI4]], [[WIDE_LOAD7]]
 ; INTERLEAVE-4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
-; INTERLEAVE-4-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; INTERLEAVE-4-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; INTERLEAVE-4-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; INTERLEAVE-4-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; INTERLEAVE-4:       middle.block:
-; INTERLEAVE-4-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP13]], [[TMP12]]
-; INTERLEAVE-4-NEXT:    [[BIN_RDX8:%.*]] = add <4 x i32> [[TMP14]], [[BIN_RDX]]
-; INTERLEAVE-4-NEXT:    [[BIN_RDX9:%.*]] = add <4 x i32> [[TMP15]], [[BIN_RDX8]]
-; INTERLEAVE-4-NEXT:    [[TMP17:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX9]])
+; INTERLEAVE-4-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP5]], [[TMP4]]
+; INTERLEAVE-4-NEXT:    [[BIN_RDX8:%.*]] = add <4 x i32> [[TMP6]], [[BIN_RDX]]
+; INTERLEAVE-4-NEXT:    [[BIN_RDX9:%.*]] = add <4 x i32> [[TMP7]], [[BIN_RDX8]]
+; INTERLEAVE-4-NEXT:    [[TMP9:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX9]])
 ; INTERLEAVE-4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
 ; INTERLEAVE-4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
 ; INTERLEAVE-4:       vec.epilog.iter.check:
@@ -56,27 +56,27 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-4-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
 ; INTERLEAVE-4:       vec.epilog.ph:
 ; INTERLEAVE-4-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
-; INTERLEAVE-4-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP17]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; INTERLEAVE-4-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP9]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
 ; INTERLEAVE-4-NEXT:    [[N_MOD_VF10:%.*]] = urem i64 [[N]], 4
 ; INTERLEAVE-4-NEXT:    [[N_VEC11:%.*]] = sub i64 [[N]], [[N_MOD_VF10]]
-; INTERLEAVE-4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[BC_MERGE_RDX]], i32 0
+; INTERLEAVE-4-NEXT:    [[TMP10:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[BC_MERGE_RDX]], i32 0
 ; INTERLEAVE-4-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
 ; INTERLEAVE-4:       vec.epilog.vector.body:
 ; INTERLEAVE-4-NEXT:    [[INDEX12:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT15:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI13:%.*]] = phi <4 x i32> [ [[TMP18]], [[VEC_EPILOG_PH]] ], [ [[TMP22:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[INDEX12]]
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD14:%.*]] = load <4 x i32>, ptr [[TMP20]], align 1
-; INTERLEAVE-4-NEXT:    [[TMP22]] = add <4 x i32> [[VEC_PHI13]], [[WIDE_LOAD14]]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI13:%.*]] = phi <4 x i32> [ [[TMP10]], [[VEC_EPILOG_PH]] ], [ [[TMP12:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[INDEX12]]
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD14:%.*]] = load <4 x i32>, ptr [[TMP11]], align 1
+; INTERLEAVE-4-NEXT:    [[TMP12]] = add <4 x i32> [[VEC_PHI13]], [[WIDE_LOAD14]]
 ; INTERLEAVE-4-NEXT:    [[INDEX_NEXT15]] = add nuw i64 [[INDEX12]], 4
-; INTERLEAVE-4-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC11]]
-; INTERLEAVE-4-NEXT:    br i1 [[TMP23]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; INTERLEAVE-4-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC11]]
+; INTERLEAVE-4-NEXT:    br i1 [[TMP13]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
 ; INTERLEAVE-4:       vec.epilog.middle.block:
-; INTERLEAVE-4-NEXT:    [[TMP24:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP22]])
+; INTERLEAVE-4-NEXT:    [[TMP14:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP12]])
 ; INTERLEAVE-4-NEXT:    [[CMP_N16:%.*]] = icmp eq i64 [[N]], [[N_VEC11]]
 ; INTERLEAVE-4-NEXT:    br i1 [[CMP_N16]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
 ; INTERLEAVE-4:       vec.epilog.scalar.ph:
 ; INTERLEAVE-4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC11]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
-; INTERLEAVE-4-NEXT:    [[BC_MERGE_RDX17:%.*]] = phi i32 [ [[TMP24]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[TMP17]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK]] ]
+; INTERLEAVE-4-NEXT:    [[BC_MERGE_RDX17:%.*]] = phi i32 [ [[TMP14]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[TMP9]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK]] ]
 ; INTERLEAVE-4-NEXT:    br label [[LOOP:%.*]]
 ; INTERLEAVE-4:       loop:
 ; INTERLEAVE-4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
@@ -88,7 +88,7 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-4-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
 ; INTERLEAVE-4-NEXT:    br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP4:![0-9]+]]
 ; INTERLEAVE-4:       exit:
-; INTERLEAVE-4-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP17]], [[MIDDLE_BLOCK]] ], [ [[TMP24]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; INTERLEAVE-4-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP9]], [[MIDDLE_BLOCK]] ], [ [[TMP14]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
 ; INTERLEAVE-4-NEXT:    ret i32 [[RED_NEXT_LCSSA]]
 ;
 ; INTERLEAVE-2-LABEL: @interleave_integer_reduction(
@@ -101,25 +101,25 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-2-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; INTERLEAVE-2:       vector.body:
 ; INTERLEAVE-2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-2-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-2-NEXT:    [[VEC_PHI1:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
-; INTERLEAVE-2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 4
-; INTERLEAVE-2-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 1
-; INTERLEAVE-2-NEXT:    [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP5]], align 1
-; INTERLEAVE-2-NEXT:    [[TMP6]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
-; INTERLEAVE-2-NEXT:    [[TMP7]] = add <4 x i32> [[VEC_PHI1]], [[WIDE_LOAD2]]
+; INTERLEAVE-2-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-2-NEXT:    [[VEC_PHI1:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-2-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
+; INTERLEAVE-2-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 4
+; INTERLEAVE-2-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 1
+; INTERLEAVE-2-NEXT:    [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP1]], align 1
+; INTERLEAVE-2-NEXT:    [[TMP2]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
+; INTERLEAVE-2-NEXT:    [[TMP3]] = add <4 x i32> [[VEC_PHI1]], [[WIDE_LOAD2]]
 ; INTERLEAVE-2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; INTERLEAVE-2-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; INTERLEAVE-2-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; INTERLEAVE-2-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; INTERLEAVE-2-NEXT:    br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; INTERLEAVE-2:       middle.block:
-; INTERLEAVE-2-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP7]], [[TMP6]]
-; INTERLEAVE-2-NEXT:    [[TMP9:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX]])
+; INTERLEAVE-2-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP3]], [[TMP2]]
+; INTERLEAVE-2-NEXT:    [[TMP5:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX]])
 ; INTERLEAVE-2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
 ; INTERLEAVE-2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; INTERLEAVE-2:       scalar.ph:
 ; INTERLEAVE-2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; INTERLEAVE-2-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP9]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; INTERLEAVE-2-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP5]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
 ; INTERLEAVE-2-NEXT:    br label [[LOOP:%.*]]
 ; INTERLEAVE-2:       loop:
 ; INTERLEAVE-2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
@@ -131,27 +131,30 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-2-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
 ; INTERLEAVE-2-NEXT:    br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
 ; INTERLEAVE-2:       exit:
-; INTERLEAVE-2-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP9]], [[MIDDLE_BLOCK]] ]
+; INTERLEAVE-2-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP5]], [[MIDDLE_BLOCK]] ]
 ; INTERLEAVE-2-NEXT:    ret i32 [[RED_NEXT_LCSSA]]
 ;
 ; INTERLEAVE-4-VLA-LABEL: @interleave_integer_reduction(
-; INTERLEAVE-4-VLA-NEXT:  entry:
-; INTERLEAVE-4-VLA-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
-; INTERLEAVE-4-VLA-NEXT:    [[TMP1:%.*]] = shl nuw i64 [[TMP0]], 4
-; INTERLEAVE-4-VLA-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], [[TMP1]]
-; INTERLEAVE-4-VLA-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; INTERLEAVE-4-VLA:       vector.ph:
+; INTERLEAVE-4-VLA-NEXT:  iter.check:
+; INTERLEAVE-4-VLA-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 2
+; INTERLEAVE-4-VLA-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]]
+; INTERLEAVE-4-VLA:       vector.main.loop.iter.check:
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
-; INTERLEAVE-4-VLA-NEXT:    [[TMP3:%.*]] = mul nuw i64 [[TMP2]], 16
-; INTERLEAVE-4-VLA-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP3:%.*]] = shl nuw i64 [[TMP2]], 4
+; INTERLEAVE-4-VLA-NEXT:    [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[N]], [[TMP3]]
+; INTERLEAVE-4-VLA-NEXT:    br i1 [[MIN_ITERS_CHECK1]], label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH:%.*]]
+; INTERLEAVE-4-VLA:       vector.ph:
+; INTERLEAVE-4-VLA-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; INTERLEAVE-4-VLA-NEXT:    [[TMP5:%.*]] = mul nuw i64 [[TMP4]], 16
+; INTERLEAVE-4-VLA-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP5]]
 ; INTERLEAVE-4-VLA-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
 ; INTERLEAVE-4-VLA-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; INTERLEAVE-4-VLA:       vector.body:
 ; INTERLEAVE-4-VLA-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; INTERLEAVE-4-VLA-NEXT:    [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP16:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-VLA-NEXT:    [[VEC_PHI1:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP17:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-VLA-NEXT:    [[VEC_PHI2:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-VLA-NEXT:    [[VEC_PHI3:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP19:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[VEC_PHI2:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP17:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[VEC_PHI3:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[VEC_PHI4:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP19:%.*]], [[VECTOR_BODY]] ]
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP8:%.*]] = shl nuw i64 [[TMP7]], 2
@@ -163,38 +166,62 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP14:%.*]] = mul nuw i64 [[TMP13]], 12
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i64 [[TMP14]]
 ; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP6]], align 1
-; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD4:%.*]] = load <vscale x 4 x i32>, ptr [[TMP9]], align 1
-; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD5:%.*]] = load <vscale x 4 x i32>, ptr [[TMP12]], align 1
-; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD6:%.*]] = load <vscale x 4 x i32>, ptr [[TMP15]], align 1
+; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD5:%.*]] = load <vscale x 4 x i32>, ptr [[TMP9]], align 1
+; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD6:%.*]] = load <vscale x 4 x i32>, ptr [[TMP12]], align 1
+; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD7:%.*]] = load <vscale x 4 x i32>, ptr [[TMP15]], align 1
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP16]] = add <vscale x 4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
-; INTERLEAVE-4-VLA-NEXT:    [[TMP17]] = add <vscale x 4 x i32> [[VEC_PHI1]], [[WIDE_LOAD4]]
-; INTERLEAVE-4-VLA-NEXT:    [[TMP18]] = add <vscale x 4 x i32> [[VEC_PHI2]], [[WIDE_LOAD5]]
-; INTERLEAVE-4-VLA-NEXT:    [[TMP19]] = add <vscale x 4 x i32> [[VEC_PHI3]], [[WIDE_LOAD6]]
-; INTERLEAVE-4-VLA-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP17]] = add <vscale x 4 x i32> [[VEC_PHI2]], [[WIDE_LOAD5]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP18]] = add <vscale x 4 x i32> [[VEC_PHI3]], [[WIDE_LOAD6]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP19]] = add <vscale x 4 x i32> [[VEC_PHI4]], [[WIDE_LOAD7]]
+; INTERLEAVE-4-VLA-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; INTERLEAVE-4-VLA-NEXT:    br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; INTERLEAVE-4-VLA:       middle.block:
 ; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX:%.*]] = add <vscale x 4 x i32> [[TMP17]], [[TMP16]]
-; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX7:%.*]] = add <vscale x 4 x i32> [[TMP18]], [[BIN_RDX]]
-; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX8:%.*]] = add <vscale x 4 x i32> [[TMP19]], [[BIN_RDX7]]
-; INTERLEAVE-4-VLA-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[BIN_RDX8]])
+; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX8:%.*]] = add <vscale x 4 x i32> [[TMP18]], [[BIN_RDX]]
+; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX9:%.*]] = add <vscale x 4 x i32> [[TMP19]], [[BIN_RDX8]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[BIN_RDX9]])
 ; INTERLEAVE-4-VLA-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
-; INTERLEAVE-4-VLA-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; INTERLEAVE-4-VLA:       scalar.ph:
-; INTERLEAVE-4-VLA-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; INTERLEAVE-4-VLA-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP21]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; INTERLEAVE-4-VLA-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
+; INTERLEAVE-4-VLA:       vec.epilog.iter.check:
+; INTERLEAVE-4-VLA-NEXT:    [[N_VEC_REMAINING:%.*]] = sub i64 [[N]], [[N_VEC]]
+; INTERLEAVE-4-VLA-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], 2
+; INTERLEAVE-4-VLA-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; INTERLEAVE-4-VLA:       vec.epilog.ph:
+; INTERLEAVE-4-VLA-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP21]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[N_MOD_VF10:%.*]] = urem i64 [[N]], 2
+; INTERLEAVE-4-VLA-NEXT:    [[N_VEC11:%.*]] = sub i64 [[N]], [[N_MOD_VF10]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP23:%.*]] = insertelement <2 x i32> zeroinitializer, i32 [[BC_MERGE_RDX]], i32 0
+; INTERLEAVE-4-VLA-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
+; INTERLEAVE-4-VLA:       vec.epilog.vector.body:
+; INTERLEAVE-4-VLA-NEXT:    [[INDEX12:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT15:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[VEC_PHI13:%.*]] = phi <2 x i32> [ [[TMP23]], [[VEC_EPILOG_PH]] ], [ [[TMP22:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[INDEX12]]
+; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD14:%.*]] = load <2 x i32>, ptr [[TMP27]], align 1
+; INTERLEAVE-4-VLA-NEXT:    [[TMP22]] = add <2 x i32> [[VEC_PHI13]], [[WIDE_LOAD14]]
+; INTERLEAVE-4-VLA-NEXT:    [[INDEX_NEXT15]] = add nuw i64 [[INDEX12]], 2
+; INTERLEAVE-4-VLA-NEXT:    [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC11]]
+; INTERLEAVE-4-VLA-NEXT:    br i1 [[TMP29]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; INTERLEAVE-4-VLA:       vec.epilog.middle.block:
+; INTERLEAVE-4-VLA-NEXT:    [[TMP30:%.*]] = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> [[TMP22]])
+; INTERLEAVE-4-VLA-NEXT:    [[CMP_N16:%.*]] = icmp eq i64 [[N]], [[N_VEC11]]
+; INTERLEAVE-4-VLA-NEXT:    br i1 [[CMP_N16]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
+; INTERLEAVE-4-VLA:       vec.epilog.scalar.ph:
+; INTERLEAVE-4-VLA-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC11]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[BC_MERGE_RDX17:%.*]] = phi i32 [ [[TMP30]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[TMP21]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK]] ]
 ; INTERLEAVE-4-VLA-NEXT:    br label [[LOOP:%.*]]
 ; INTERLEAVE-4-VLA:       loop:
-; INTERLEAVE-4-VLA-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; INTERLEAVE-4-VLA-NEXT:    [[RED:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[RED_NEXT:%.*]], [[LOOP]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[RED:%.*]] = phi i32 [ [[BC_MERGE_RDX17]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[RED_NEXT:%.*]], [[LOOP]] ]
 ; INTERLEAVE-4-VLA-NEXT:    [[GEP_SRC:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[IV]]
 ; INTERLEAVE-4-VLA-NEXT:    [[L:%.*]] = load i32, ptr [[GEP_SRC]], align 1
 ; INTERLEAVE-4-VLA-NEXT:    [[RED_NEXT]] = add i32 [[RED]], [[L]]
 ; INTERLEAVE-4-VLA-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
 ; INTERLEAVE-4-VLA-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; INTERLEAVE-4-VLA-NEXT:    br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
+; INTERLEAVE-4-VLA-NEXT:    br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP4:![0-9]+]]
 ; INTERLEAVE-4-VLA:       exit:
-; INTERLEAVE-4-VLA-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP21]], [[MIDDLE_BLOCK]] ]
+; INTERLEAVE-4-VLA-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP21]], [[MIDDLE_BLOCK]] ], [ [[TMP30]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
 ; INTERLEAVE-4-VLA-NEXT:    ret i32 [[RED_NEXT_LCSSA]]
 ;
 entry:
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll b/llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll
index 8b2da8c4a7047..55e24486de4b6 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll
@@ -376,12 +376,17 @@ define i16 @reduce_udiv(ptr %src, i16 %x, i64 %N) #0 {
 ;
 ; VSCALEFORTUNING2-LABEL: define i16 @reduce_udiv(
 ; VSCALEFORTUNING2-SAME: ptr [[SRC:%.*]], i16 [[X:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
-; VSCALEFORTUNING2-NEXT:  [[ENTRY:.*]]:
+; VSCALEFORTUNING2-NEXT:  [[ITER_CHECK:.*]]:
 ; VSCALEFORTUNING2-NEXT:    [[TMP0:%.*]] = add i64 [[N]], 1
 ; VSCALEFORTUNING2-NEXT:    [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
-; VSCALEFORTUNING2-NEXT:    [[TMP2:%.*]] = shl nuw i64 [[TMP1]], 3
+; VSCALEFORTUNING2-NEXT:    [[TMP5:%.*]] = shl nuw i64 [[TMP1]], 1
+; VSCALEFORTUNING2-NEXT:    [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[TMP0]], [[TMP5]]
+; VSCALEFORTUNING2-NEXT:    br i1 [[MIN_ITERS_CHECK1]], label %[[VEC_EPILOG_SCALAR_PH:.*]], label %[[VECTOR_MAIN_LOOP_ITER_CHECK:.*]]
+; VSCALEFORTUNING2:       [[VECTOR_MAIN_LOOP_ITER_CHECK]]:
+; VSCALEFORTUNING2-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; VSCALEFORTUNING2-NEXT:    [[TMP2:%.*]] = shl nuw i64 [[TMP6]], 3
 ; VSCALEFORTUNING2-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], [[TMP2]]
-; VSCALEFORTUNING2-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; VSCALEFORTUNING2-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[VEC_EPILOG_PH:.*]], label %[[VECTOR_PH:.*]]
 ; VSCALEFORTUNING2:       [[VECTOR_PH]]:
 ; VSCALEFORTUNING2-NEXT:    [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
 ; VSCALEFORTUNING2-NEXT:    [[TMP4:%.*]] = mul nuw i64 [[TMP3]], 8
@@ -411,23 +416,54 @@ define i16 @reduce_udiv(ptr %src, i16 %x, i64 %N) #0 {
 ; VSCALEFORTUNING2-NEXT:    [[BIN_RDX:%.*]] = or <vscale x 4 x i16> [[TMP16]], [[TMP15]]
 ; VSCALEFORTUNING2-NEXT:    [[TMP18:%.*]] = call i16 @llvm.vector.reduce.or.nxv4i16(<vscale x 4 x i16> [[BIN_RDX]])
 ; VSCALEFORTUNING2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
-; VSCALEFORTUNING2-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
-; VSCALEFORTUNING2:       [[SCALAR_PH]]:
-; VSCALEFORTUNING2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
-; VSCALEFORTUNING2-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i16 [ [[TMP18]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
+; VSCALEFORTUNING2-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[VEC_EPILOG_ITER_CHECK:.*]]
+; VSCALEFORTUNING2:       [[VEC_EPILOG_ITER_CHECK]]:
+; VSCALEFORTUNING2-NEXT:    [[N_VEC_REMAINING:%.*]] = sub i64 [[TMP0]], [[N_VEC]]
+; VSCALEFORTUNING2-NEXT:    [[TMP22:%.*]] = call i64 @llvm.vscale.i64()
+; VSCALEFORTUNING2-NEXT:    [[TMP27:%.*]] = shl nuw i64 [[TMP22]], 1
+; VSCALEFORTUNING2-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], [[TMP27]]
+; VSCALEFORTUNING2-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VEC_EPILOG_PH]]
+; VSCALEFORTUNING2:       [[VEC_EPILOG_PH]]:
+; VSCALEFORTUNING2-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; VSCALEFORTUNING2-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i16 [ [[TMP18]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; VSCALEFORTUNING2-NEXT:    [[TMP19:%.*]] = call i64 @llvm.vscale.i64()
+; VSCALEFORTUNING2-NEXT:    [[TMP20:%.*]] = mul nuw i64 [[TMP19]], 2
+; VSCALEFORTUNING2-NEXT:    [[N_MOD_VF4:%.*]] = urem i64 [[TMP0]], [[TMP20]]
+; VSCALEFORTUNING2-NEXT:    [[N_VEC5:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF4]]
+; VSCALEFORTUNING2-NEXT:    [[TMP21:%.*]] = insertelement <vscale x 2 x i16> zeroinitializer, i16 [[BC_MERGE_RDX]], i32 0
+; VSCALEFORTUNING2-NEXT:    [[BROADCAST_SPLATINSERT6:%.*]] = insertelement <vscale x 2 x i16> poison, i16 [[X]], i64 0
+; VSCALEFORTUNING2-NEXT:    [[BROADCAST_SPLAT7:%.*]] = shufflevector <vscale x 2 x i16> [[BROADCAST_SPLATINSERT6]], <vscale x 2 x i16> poison, <vscale x 2 x i32> zeroinitializer
+; VSCALEFORTUNING2-NEXT:    br label %[[VEC_EPILOG_VECTOR_BODY:.*]]
+; VSCALEFORTUNING2:       [[VEC_EPILOG_VECTOR_BODY]]:
+; VSCALEFORTUNING2-NEXT:    [[IV:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT11:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
+; VSCALEFORTUNING2-NEXT:    [[VEC_PHI9:%.*]] = phi <vscale x 2 x i16> [ [[TMP21]], %[[VEC_EPILOG_PH]] ], [ [[TMP24:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
+; VSCALEFORTUNING2-NEXT:    [[GEP:%.*]] = getelementptr i16, ptr [[SRC]], i64 [[IV]]
+; VSCALEFORTUNING2-NEXT:    [[WIDE_LOAD10:%.*]] = load <vscale x 2 x i16>, ptr [[GEP]], align 2
+; VSCALEFORTUNING2-NEXT:    [[TMP23:%.*]] = udiv <vscale x 2 x i16> [[WIDE_LOAD10]], [[BROADCAST_SPLAT7]]
+; VSCALEFORTUNING2-NEXT:    [[TMP24]] = or <vscale x 2 x i16> [[TMP23]], [[VEC_PHI9]]
+; VSCALEFORTUNING2-NEXT:    [[INDEX_NEXT11]] = add nuw i64 [[IV]], [[TMP20]]
+; VSCALEFORTUNING2-NEXT:    [[TMP25:%.*]] = icmp eq i64 [[INDEX_NEXT11]], [[N_VEC5]]
+; VSCALEFORTUNING2-NEXT:    br i1 [[TMP25]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; VSCALEFORTUNING2:       [[VEC_EPILOG_MIDDLE_BLOCK]]:
+; VSCALEFORTUNING2-NEXT:    [[TMP26:%.*]] = call i16 @llvm.vector.reduce.or.nxv2i16(<vscale x 2 x i16> [[TMP24]])
+; VSCALEFORTUNING2-NEXT:    [[CMP_N12:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC5]]
+; VSCALEFORTUNING2-NEXT:    br i1 [[CMP_N12]], label %[[EXIT]], label %[[VEC_EPILOG_SCALAR_PH]]
+; VSCALEFORTUNING2:       [[VEC_EPILOG_SCALAR_PH]]:
+; VSCALEFORTUNING2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC5]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[ITER_CHECK]] ]
+; VSCALEFORTUNING2-NEXT:    [[BC_MERGE_RDX13:%.*]] = phi i16 [ [[TMP26]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[TMP18]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[ITER_CHECK]] ]
 ; VSCALEFORTUNING2-NEXT:    br label %[[LOOP:.*]]
 ; VSCALEFORTUNING2:       [[LOOP]]:
-; VSCALEFORTUNING2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
-; VSCALEFORTUNING2-NEXT:    [[RED:%.*]] = phi i16 [ [[BC_MERGE_RDX]], %[[SCALAR_PH]] ], [ [[RED_NEXT:%.*]], %[[LOOP]] ]
-; VSCALEFORTUNING2-NEXT:    [[GEP:%.*]] = getelementptr i16, ptr [[SRC]], i64 [[IV]]
-; VSCALEFORTUNING2-NEXT:    [[L:%.*]] = load i16, ptr [[GEP]], align 2
+; VSCALEFORTUNING2-NEXT:    [[IV1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
+; VSCALEFORTUNING2-NEXT:    [[RED:%.*]] = phi i16 [ [[BC_MERGE_RDX13]], %[[VEC_EPILOG_SCALAR_PH]] ], [ [[RED_NEXT:%.*]], %[[LOOP]] ]
+; VSCALEFORTUNING2-NEXT:    [[GEP1:%.*]] = getelementptr i16, ptr [[SRC]], i64 [[IV1]]
+; VSCALEFORTUNING2-NEXT:    [[L:%.*]] = load i16, ptr [[GEP1]], align 2
 ; VSCALEFORTUNING2-NEXT:    [[DIV:%.*]] = udiv i16 [[L]], [[X]]
 ; VSCALEFORTUNING2-NEXT:    [[RED_NEXT]] = or i16 [[DIV]], [[RED]]
-; VSCALEFORTUNING2-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
-; VSCALEFORTUNING2-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV]], [[N]]
-; VSCALEFORTUNING2-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
+; VSCALEFORTUNING2-NEXT:    [[IV_NEXT]] = add i64 [[IV1]], 1
+; VSCALEFORTUNING2-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV1]], [[N]]
+; VSCALEFORTUNING2-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP6:![0-9]+]]
 ; VSCALEFORTUNING2:       [[EXIT]]:
-; VSCALEFORTUNING2-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i16 [ [[RED_NEXT]], %[[LOOP]] ], [ [[TMP18]], %[[MIDDLE_BLOCK]] ]
+; VSCALEFORTUNING2-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i16 [ [[RED_NEXT]], %[[LOOP]] ], [ [[TMP18]], %[[MIDDLE_BLOCK]] ], [ [[TMP26]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ]
 ; VSCALEFORTUNING2-NEXT:    ret i16 [[RED_NEXT_LCSSA]]
 ;
 ; PRED-LABEL: define i16 @reduce_udiv(

>From c892779741d58d8130dcb019f9c2bf977d10b6af Mon Sep 17 00:00:00 2001
From: Hassnaa Hamdi <hasham01 at e140134.arm.com>
Date: Sat, 6 Sep 2025 19:37:27 +0100
Subject: [PATCH 2/3] rebase

---
 .../AArch64/induction-costs-sve.ll            | 321 +++++++++---------
 1 file changed, 160 insertions(+), 161 deletions(-)

diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll b/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
index 02b1fc34f7fc6..ffac7e6faeaf4 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
@@ -24,80 +24,79 @@ define void @iv_casts(ptr %dst, ptr %src, i32 %x, i64 %N) #0 {
 ; DEFAULT-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP6]], [[TMP5]]
 ; DEFAULT-NEXT:    br i1 [[DIFF_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VECTOR_MAIN_LOOP_ITER_CHECK:.*]]
 ; DEFAULT:       [[VECTOR_MAIN_LOOP_ITER_CHECK]]:
-; DEFAULT-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
-; DEFAULT-NEXT:    [[TMP8:%.*]] = shl nuw i64 [[TMP9]], 4
+; DEFAULT-NEXT:    [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP8:%.*]] = shl nuw i64 [[TMP7]], 4
 ; DEFAULT-NEXT:    [[MIN_ITERS_CHECK3:%.*]] = icmp ult i64 [[TMP0]], [[TMP8]]
 ; DEFAULT-NEXT:    br i1 [[MIN_ITERS_CHECK3]], label %[[VEC_EPILOG_PH:.*]], label %[[VECTOR_PH:.*]]
 ; DEFAULT:       [[VECTOR_PH]]:
-; DEFAULT-NEXT:    [[TMP11:%.*]] = call i64 @llvm.vscale.i64()
-; DEFAULT-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP11]], 16
+; DEFAULT-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP9]], 16
 ; DEFAULT-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], [[TMP10]]
 ; DEFAULT-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
 ; DEFAULT-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 8 x i32> poison, i32 [[X]], i64 0
 ; DEFAULT-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 8 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
-; DEFAULT-NEXT:    [[TMP13:%.*]] = trunc <vscale x 8 x i32> [[BROADCAST_SPLAT]] to <vscale x 8 x i16>
+; DEFAULT-NEXT:    [[TMP11:%.*]] = trunc <vscale x 8 x i32> [[BROADCAST_SPLAT]] to <vscale x 8 x i16>
 ; DEFAULT-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; DEFAULT:       [[VECTOR_BODY]]:
 ; DEFAULT-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; DEFAULT-NEXT:    [[TMP20:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[INDEX]]
-; DEFAULT-NEXT:    [[TMP23:%.*]] = call i64 @llvm.vscale.i64()
-; DEFAULT-NEXT:    [[TMP24:%.*]] = shl nuw i64 [[TMP23]], 3
-; DEFAULT-NEXT:    [[TMP25:%.*]] = getelementptr i8, ptr [[TMP20]], i64 [[TMP24]]
-; DEFAULT-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 8 x i8>, ptr [[TMP20]], align 1
-; DEFAULT-NEXT:    [[WIDE_LOAD4:%.*]] = load <vscale x 8 x i8>, ptr [[TMP25]], align 1
-; DEFAULT-NEXT:    [[TMP26:%.*]] = zext <vscale x 8 x i8> [[WIDE_LOAD]] to <vscale x 8 x i16>
-; DEFAULT-NEXT:    [[TMP27:%.*]] = zext <vscale x 8 x i8> [[WIDE_LOAD4]] to <vscale x 8 x i16>
-; DEFAULT-NEXT:    [[TMP28:%.*]] = mul <vscale x 8 x i16> [[TMP26]], [[TMP13]]
-; DEFAULT-NEXT:    [[TMP29:%.*]] = mul <vscale x 8 x i16> [[TMP27]], [[TMP13]]
-; DEFAULT-NEXT:    [[TMP32:%.*]] = or <vscale x 8 x i16> [[TMP28]], [[TMP26]]
-; DEFAULT-NEXT:    [[TMP33:%.*]] = or <vscale x 8 x i16> [[TMP29]], [[TMP27]]
-; DEFAULT-NEXT:    [[TMP34:%.*]] = lshr <vscale x 8 x i16> [[TMP32]], splat (i16 1)
-; DEFAULT-NEXT:    [[TMP35:%.*]] = lshr <vscale x 8 x i16> [[TMP33]], splat (i16 1)
-; DEFAULT-NEXT:    [[TMP36:%.*]] = trunc <vscale x 8 x i16> [[TMP34]] to <vscale x 8 x i8>
-; DEFAULT-NEXT:    [[TMP37:%.*]] = trunc <vscale x 8 x i16> [[TMP35]] to <vscale x 8 x i8>
-; DEFAULT-NEXT:    [[TMP38:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX]]
-; DEFAULT-NEXT:    [[TMP41:%.*]] = call i64 @llvm.vscale.i64()
-; DEFAULT-NEXT:    [[TMP42:%.*]] = shl nuw i64 [[TMP41]], 3
-; DEFAULT-NEXT:    [[TMP43:%.*]] = getelementptr i8, ptr [[TMP38]], i64 [[TMP42]]
-; DEFAULT-NEXT:    store <vscale x 8 x i8> [[TMP36]], ptr [[TMP38]], align 1
-; DEFAULT-NEXT:    store <vscale x 8 x i8> [[TMP37]], ptr [[TMP43]], align 1
+; DEFAULT-NEXT:    [[TMP12:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[INDEX]]
+; DEFAULT-NEXT:    [[TMP13:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP14:%.*]] = shl nuw i64 [[TMP13]], 3
+; DEFAULT-NEXT:    [[TMP15:%.*]] = getelementptr i8, ptr [[TMP12]], i64 [[TMP14]]
+; DEFAULT-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 8 x i8>, ptr [[TMP12]], align 1
+; DEFAULT-NEXT:    [[WIDE_LOAD4:%.*]] = load <vscale x 8 x i8>, ptr [[TMP15]], align 1
+; DEFAULT-NEXT:    [[TMP16:%.*]] = zext <vscale x 8 x i8> [[WIDE_LOAD]] to <vscale x 8 x i16>
+; DEFAULT-NEXT:    [[TMP17:%.*]] = zext <vscale x 8 x i8> [[WIDE_LOAD4]] to <vscale x 8 x i16>
+; DEFAULT-NEXT:    [[TMP18:%.*]] = mul <vscale x 8 x i16> [[TMP16]], [[TMP11]]
+; DEFAULT-NEXT:    [[TMP19:%.*]] = mul <vscale x 8 x i16> [[TMP17]], [[TMP11]]
+; DEFAULT-NEXT:    [[TMP20:%.*]] = or <vscale x 8 x i16> [[TMP18]], [[TMP16]]
+; DEFAULT-NEXT:    [[TMP21:%.*]] = or <vscale x 8 x i16> [[TMP19]], [[TMP17]]
+; DEFAULT-NEXT:    [[TMP22:%.*]] = lshr <vscale x 8 x i16> [[TMP20]], splat (i16 1)
+; DEFAULT-NEXT:    [[TMP23:%.*]] = lshr <vscale x 8 x i16> [[TMP21]], splat (i16 1)
+; DEFAULT-NEXT:    [[TMP24:%.*]] = trunc <vscale x 8 x i16> [[TMP22]] to <vscale x 8 x i8>
+; DEFAULT-NEXT:    [[TMP25:%.*]] = trunc <vscale x 8 x i16> [[TMP23]] to <vscale x 8 x i8>
+; DEFAULT-NEXT:    [[TMP26:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX]]
+; DEFAULT-NEXT:    [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP28:%.*]] = shl nuw i64 [[TMP27]], 3
+; DEFAULT-NEXT:    [[TMP29:%.*]] = getelementptr i8, ptr [[TMP26]], i64 [[TMP28]]
+; DEFAULT-NEXT:    store <vscale x 8 x i8> [[TMP24]], ptr [[TMP26]], align 1
+; DEFAULT-NEXT:    store <vscale x 8 x i8> [[TMP25]], ptr [[TMP29]], align 1
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP10]]
-; DEFAULT-NEXT:    [[TMP44:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP44]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; DEFAULT-NEXT:    [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; DEFAULT-NEXT:    br i1 [[TMP30]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[VEC_EPILOG_ITER_CHECK:.*]]
 ; DEFAULT:       [[VEC_EPILOG_ITER_CHECK]]:
 ; DEFAULT-NEXT:    [[N_VEC_REMAINING:%.*]] = sub i64 [[TMP0]], [[N_VEC]]
-; DEFAULT-NEXT:    [[TMP47:%.*]] = call i64 @llvm.vscale.i64()
-; DEFAULT-NEXT:    [[TMP48:%.*]] = shl nuw i64 [[TMP47]], 2
-; DEFAULT-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], [[TMP48]]
-; DEFAULT-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VEC_EPILOG_PH]]
+; DEFAULT-NEXT:    [[TMP31:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP32:%.*]] = shl nuw i64 [[TMP31]], 2
+; DEFAULT-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], [[TMP32]]
+; DEFAULT-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VEC_EPILOG_PH]], !prof [[PROF3:![0-9]+]]
 ; DEFAULT:       [[VEC_EPILOG_PH]]:
 ; DEFAULT-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
-; DEFAULT-NEXT:    [[TMP49:%.*]] = call i64 @llvm.vscale.i64()
-; DEFAULT-NEXT:    [[TMP50:%.*]] = mul nuw i64 [[TMP49]], 4
-; DEFAULT-NEXT:    [[N_MOD_VF5:%.*]] = urem i64 [[TMP0]], [[TMP50]]
+; DEFAULT-NEXT:    [[TMP33:%.*]] = call i64 @llvm.vscale.i64()
+; DEFAULT-NEXT:    [[TMP34:%.*]] = mul nuw i64 [[TMP33]], 4
+; DEFAULT-NEXT:    [[N_MOD_VF5:%.*]] = urem i64 [[TMP0]], [[TMP34]]
 ; DEFAULT-NEXT:    [[N_VEC6:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF5]]
 ; DEFAULT-NEXT:    [[BROADCAST_SPLATINSERT7:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[X]], i64 0
 ; DEFAULT-NEXT:    [[BROADCAST_SPLAT8:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT7]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
-; DEFAULT-NEXT:    [[TMP51:%.*]] = trunc <vscale x 4 x i32> [[BROADCAST_SPLAT8]] to <vscale x 4 x i16>
+; DEFAULT-NEXT:    [[TMP35:%.*]] = trunc <vscale x 4 x i32> [[BROADCAST_SPLAT8]] to <vscale x 4 x i16>
 ; DEFAULT-NEXT:    br label %[[VEC_EPILOG_VECTOR_BODY:.*]]
 ; DEFAULT:       [[VEC_EPILOG_VECTOR_BODY]]:
-; DEFAULT-NEXT:    [[IV:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT11:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
-; DEFAULT-NEXT:    [[GEP_SRC:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[IV]]
-; DEFAULT-NEXT:    [[WIDE_LOAD10:%.*]] = load <vscale x 4 x i8>, ptr [[GEP_SRC]], align 1
-; DEFAULT-NEXT:    [[TMP39:%.*]] = zext <vscale x 4 x i8> [[WIDE_LOAD10]] to <vscale x 4 x i16>
-; DEFAULT-NEXT:    [[TMP40:%.*]] = mul <vscale x 4 x i16> [[TMP39]], [[TMP51]]
-; DEFAULT-NEXT:    [[TMP52:%.*]] = zext <vscale x 4 x i8> [[WIDE_LOAD10]] to <vscale x 4 x i16>
-; DEFAULT-NEXT:    [[TMP53:%.*]] = or <vscale x 4 x i16> [[TMP40]], [[TMP52]]
-; DEFAULT-NEXT:    [[TMP54:%.*]] = lshr <vscale x 4 x i16> [[TMP53]], splat (i16 1)
-; DEFAULT-NEXT:    [[TMP55:%.*]] = trunc <vscale x 4 x i16> [[TMP54]] to <vscale x 4 x i8>
-; DEFAULT-NEXT:    [[TMP45:%.*]] = getelementptr i8, ptr [[DST]], i64 [[IV]]
-; DEFAULT-NEXT:    store <vscale x 4 x i8> [[TMP55]], ptr [[TMP45]], align 1
-; DEFAULT-NEXT:    [[INDEX_NEXT11]] = add nuw i64 [[IV]], [[TMP50]]
-; DEFAULT-NEXT:    [[TMP46:%.*]] = icmp eq i64 [[INDEX_NEXT11]], [[N_VEC6]]
-; DEFAULT-NEXT:    br i1 [[TMP46]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; DEFAULT-NEXT:    [[INDEX9:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT11:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
+; DEFAULT-NEXT:    [[TMP36:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[INDEX9]]
+; DEFAULT-NEXT:    [[WIDE_LOAD10:%.*]] = load <vscale x 4 x i8>, ptr [[TMP36]], align 1
+; DEFAULT-NEXT:    [[TMP37:%.*]] = zext <vscale x 4 x i8> [[WIDE_LOAD10]] to <vscale x 4 x i16>
+; DEFAULT-NEXT:    [[TMP38:%.*]] = mul <vscale x 4 x i16> [[TMP37]], [[TMP35]]
+; DEFAULT-NEXT:    [[TMP39:%.*]] = or <vscale x 4 x i16> [[TMP38]], [[TMP37]]
+; DEFAULT-NEXT:    [[TMP40:%.*]] = lshr <vscale x 4 x i16> [[TMP39]], splat (i16 1)
+; DEFAULT-NEXT:    [[TMP41:%.*]] = trunc <vscale x 4 x i16> [[TMP40]] to <vscale x 4 x i8>
+; DEFAULT-NEXT:    [[TMP42:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX9]]
+; DEFAULT-NEXT:    store <vscale x 4 x i8> [[TMP41]], ptr [[TMP42]], align 1
+; DEFAULT-NEXT:    [[INDEX_NEXT11]] = add nuw i64 [[INDEX9]], [[TMP34]]
+; DEFAULT-NEXT:    [[TMP43:%.*]] = icmp eq i64 [[INDEX_NEXT11]], [[N_VEC6]]
+; DEFAULT-NEXT:    br i1 [[TMP43]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; DEFAULT:       [[VEC_EPILOG_MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N12:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC6]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N12]], label %[[EXIT]], label %[[VEC_EPILOG_SCALAR_PH]]
@@ -105,20 +104,20 @@ define void @iv_casts(ptr %dst, ptr %src, i32 %x, i64 %N) #0 {
 ; DEFAULT-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC6]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MEMCHECK]] ], [ 0, %[[ITER_CHECK]] ]
 ; DEFAULT-NEXT:    br label %[[LOOP:.*]]
 ; DEFAULT:       [[LOOP]]:
-; DEFAULT-NEXT:    [[IV1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
-; DEFAULT-NEXT:    [[GEP_SRC1:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[IV1]]
-; DEFAULT-NEXT:    [[L:%.*]] = load i8, ptr [[GEP_SRC1]], align 1
+; DEFAULT-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
+; DEFAULT-NEXT:    [[GEP_SRC:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[IV]]
+; DEFAULT-NEXT:    [[L:%.*]] = load i8, ptr [[GEP_SRC]], align 1
 ; DEFAULT-NEXT:    [[L_EXT:%.*]] = zext i8 [[L]] to i32
-; DEFAULT-NEXT:    [[MUL16_US:%.*]] = mul i32 [[L_EXT]], [[X]]
-; DEFAULT-NEXT:    [[IV_NEXT]] = add i64 [[IV1]], 1
-; DEFAULT-NEXT:    [[CONV25_US:%.*]] = zext i8 [[L]] to i32
-; DEFAULT-NEXT:    [[ADD34_US:%.*]] = or i32 [[MUL16_US]], [[CONV25_US]]
-; DEFAULT-NEXT:    [[SHR35_US:%.*]] = lshr i32 [[ADD34_US]], 1
-; DEFAULT-NEXT:    [[CONV36_US:%.*]] = trunc i32 [[SHR35_US]] to i8
-; DEFAULT-NEXT:    [[GEP_DST:%.*]] = getelementptr i8, ptr [[DST]], i64 [[IV1]]
-; DEFAULT-NEXT:    store i8 [[CONV36_US]], ptr [[GEP_DST]], align 1
-; DEFAULT-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV1]], [[N]]
-; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP4:![0-9]+]]
+; DEFAULT-NEXT:    [[MUL:%.*]] = mul i32 [[L_EXT]], [[X]]
+; DEFAULT-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
+; DEFAULT-NEXT:    [[L_EXT_2:%.*]] = zext i8 [[L]] to i32
+; DEFAULT-NEXT:    [[OR:%.*]] = or i32 [[MUL]], [[L_EXT_2]]
+; DEFAULT-NEXT:    [[LSHR:%.*]] = lshr i32 [[OR]], 1
+; DEFAULT-NEXT:    [[TRUNC:%.*]] = trunc i32 [[LSHR]] to i8
+; DEFAULT-NEXT:    [[GEP_DST:%.*]] = getelementptr i8, ptr [[DST]], i64 [[IV]]
+; DEFAULT-NEXT:    store i8 [[TRUNC]], ptr [[GEP_DST]], align 1
+; DEFAULT-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV]], [[N]]
+; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -140,31 +139,31 @@ define void @iv_casts(ptr %dst, ptr %src, i32 %x, i64 %N) #0 {
 ; PRED-NEXT:    [[TMP5:%.*]] = mul nuw i64 [[TMP4]], 16
 ; PRED-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 16 x i32> poison, i32 [[X]], i64 0
 ; PRED-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 16 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 16 x i32> poison, <vscale x 16 x i32> zeroinitializer
-; PRED-NEXT:    [[TMP11:%.*]] = call i64 @llvm.vscale.i64()
-; PRED-NEXT:    [[TMP12:%.*]] = shl nuw i64 [[TMP11]], 4
-; PRED-NEXT:    [[TMP13:%.*]] = sub i64 [[TMP0]], [[TMP12]]
-; PRED-NEXT:    [[TMP14:%.*]] = icmp ugt i64 [[TMP0]], [[TMP12]]
-; PRED-NEXT:    [[TMP15:%.*]] = select i1 [[TMP14]], i64 [[TMP13]], i64 0
+; PRED-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; PRED-NEXT:    [[TMP7:%.*]] = shl nuw i64 [[TMP6]], 4
+; PRED-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP0]], [[TMP7]]
+; PRED-NEXT:    [[TMP9:%.*]] = icmp ugt i64 [[TMP0]], [[TMP7]]
+; PRED-NEXT:    [[TMP10:%.*]] = select i1 [[TMP9]], i64 [[TMP8]], i64 0
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK_ENTRY:%.*]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 0, i64 [[TMP0]])
-; PRED-NEXT:    [[TMP16:%.*]] = trunc <vscale x 16 x i32> [[BROADCAST_SPLAT]] to <vscale x 16 x i16>
+; PRED-NEXT:    [[TMP11:%.*]] = trunc <vscale x 16 x i32> [[BROADCAST_SPLAT]] to <vscale x 16 x i16>
 ; PRED-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; PRED:       [[VECTOR_BODY]]:
 ; PRED-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
 ; PRED-NEXT:    [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 16 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], %[[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; PRED-NEXT:    [[TMP18:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[INDEX]]
-; PRED-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 16 x i8> @llvm.masked.load.nxv16i8.p0(ptr [[TMP18]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]], <vscale x 16 x i8> poison)
-; PRED-NEXT:    [[TMP17:%.*]] = zext <vscale x 16 x i8> [[WIDE_MASKED_LOAD]] to <vscale x 16 x i16>
-; PRED-NEXT:    [[TMP22:%.*]] = mul <vscale x 16 x i16> [[TMP17]], [[TMP16]]
-; PRED-NEXT:    [[TMP20:%.*]] = or <vscale x 16 x i16> [[TMP22]], [[TMP17]]
-; PRED-NEXT:    [[TMP21:%.*]] = lshr <vscale x 16 x i16> [[TMP20]], splat (i16 1)
-; PRED-NEXT:    [[TMP23:%.*]] = trunc <vscale x 16 x i16> [[TMP21]] to <vscale x 16 x i8>
-; PRED-NEXT:    [[TMP26:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX]]
-; PRED-NEXT:    call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> [[TMP23]], ptr [[TMP26]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
+; PRED-NEXT:    [[TMP12:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[INDEX]]
+; PRED-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 16 x i8> @llvm.masked.load.nxv16i8.p0(ptr [[TMP12]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]], <vscale x 16 x i8> poison)
+; PRED-NEXT:    [[TMP13:%.*]] = zext <vscale x 16 x i8> [[WIDE_MASKED_LOAD]] to <vscale x 16 x i16>
+; PRED-NEXT:    [[TMP14:%.*]] = mul <vscale x 16 x i16> [[TMP13]], [[TMP11]]
+; PRED-NEXT:    [[TMP15:%.*]] = or <vscale x 16 x i16> [[TMP14]], [[TMP13]]
+; PRED-NEXT:    [[TMP16:%.*]] = lshr <vscale x 16 x i16> [[TMP15]], splat (i16 1)
+; PRED-NEXT:    [[TMP17:%.*]] = trunc <vscale x 16 x i16> [[TMP16]] to <vscale x 16 x i8>
+; PRED-NEXT:    [[TMP18:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX]]
+; PRED-NEXT:    call void @llvm.masked.store.nxv16i8.p0(<vscale x 16 x i8> [[TMP17]], ptr [[TMP18]], i32 1, <vscale x 16 x i1> [[ACTIVE_LANE_MASK]])
 ; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP5]]
-; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP15]])
-; PRED-NEXT:    [[TMP25:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
-; PRED-NEXT:    [[TMP27:%.*]] = xor i1 [[TMP25]], true
-; PRED-NEXT:    br i1 [[TMP27]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; PRED-NEXT:    [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 [[INDEX]], i64 [[TMP10]])
+; PRED-NEXT:    [[TMP19:%.*]] = extractelement <vscale x 16 x i1> [[ACTIVE_LANE_MASK_NEXT]], i32 0
+; PRED-NEXT:    [[TMP20:%.*]] = xor i1 [[TMP19]], true
+; PRED-NEXT:    br i1 [[TMP20]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br label %[[EXIT:.*]]
 ; PRED:       [[SCALAR_PH]]:
@@ -175,14 +174,14 @@ define void @iv_casts(ptr %dst, ptr %src, i32 %x, i64 %N) #0 {
 ; PRED-NEXT:    [[GEP_SRC:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[IV]]
 ; PRED-NEXT:    [[L:%.*]] = load i8, ptr [[GEP_SRC]], align 1
 ; PRED-NEXT:    [[L_EXT:%.*]] = zext i8 [[L]] to i32
-; PRED-NEXT:    [[MUL16_US:%.*]] = mul i32 [[L_EXT]], [[X]]
+; PRED-NEXT:    [[MUL:%.*]] = mul i32 [[L_EXT]], [[X]]
 ; PRED-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
-; PRED-NEXT:    [[CONV25_US:%.*]] = zext i8 [[L]] to i32
-; PRED-NEXT:    [[ADD34_US:%.*]] = or i32 [[MUL16_US]], [[CONV25_US]]
-; PRED-NEXT:    [[SHR35_US:%.*]] = lshr i32 [[ADD34_US]], 1
-; PRED-NEXT:    [[CONV36_US:%.*]] = trunc i32 [[SHR35_US]] to i8
+; PRED-NEXT:    [[L_EXT_2:%.*]] = zext i8 [[L]] to i32
+; PRED-NEXT:    [[OR:%.*]] = or i32 [[MUL]], [[L_EXT_2]]
+; PRED-NEXT:    [[LSHR:%.*]] = lshr i32 [[OR]], 1
+; PRED-NEXT:    [[TRUNC:%.*]] = trunc i32 [[LSHR]] to i8
 ; PRED-NEXT:    [[GEP_DST:%.*]] = getelementptr i8, ptr [[DST]], i64 [[IV]]
-; PRED-NEXT:    store i8 [[CONV36_US]], ptr [[GEP_DST]], align 1
+; PRED-NEXT:    store i8 [[TRUNC]], ptr [[GEP_DST]], align 1
 ; PRED-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV]], [[N]]
 ; PRED-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
 ; PRED:       [[EXIT]]:
@@ -243,18 +242,18 @@ define void @iv_trunc(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT:       [[VECTOR_BODY]]:
 ; DEFAULT-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
 ; DEFAULT-NEXT:    [[TMP13:%.*]] = trunc i64 [[INDEX]] to i32
-; DEFAULT-NEXT:    [[TMP15:%.*]] = add i32 [[TMP13]], 1
-; DEFAULT-NEXT:    [[TMP16:%.*]] = mul i32 [[MUL_X]], [[TMP13]]
-; DEFAULT-NEXT:    [[TMP17:%.*]] = mul i32 [[MUL_X]], [[TMP15]]
+; DEFAULT-NEXT:    [[TMP14:%.*]] = add i32 [[TMP13]], 1
+; DEFAULT-NEXT:    [[TMP15:%.*]] = mul i32 [[MUL_X]], [[TMP13]]
+; DEFAULT-NEXT:    [[TMP16:%.*]] = mul i32 [[MUL_X]], [[TMP14]]
+; DEFAULT-NEXT:    [[TMP17:%.*]] = zext i32 [[TMP15]] to i64
 ; DEFAULT-NEXT:    [[TMP18:%.*]] = zext i32 [[TMP16]] to i64
-; DEFAULT-NEXT:    [[TMP19:%.*]] = zext i32 [[TMP17]] to i64
+; DEFAULT-NEXT:    [[TMP19:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP17]]
 ; DEFAULT-NEXT:    [[TMP20:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP18]]
-; DEFAULT-NEXT:    [[TMP21:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP19]]
+; DEFAULT-NEXT:    store i32 1, ptr [[TMP19]], align 4
 ; DEFAULT-NEXT:    store i32 1, ptr [[TMP20]], align 4
-; DEFAULT-NEXT:    store i32 1, ptr [[TMP21]], align 4
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; DEFAULT-NEXT:    [[TMP22:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP22]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; DEFAULT-NEXT:    [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; DEFAULT-NEXT:    br i1 [[TMP21]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -270,7 +269,7 @@ define void @iv_trunc(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 1, ptr [[GEP]], align 4
 ; DEFAULT-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
 ; DEFAULT-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV]], [[N]]
-; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -400,31 +399,31 @@ define void @trunc_ivs_and_store(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT:       [[VECTOR_PH]]:
 ; DEFAULT-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 2
 ; DEFAULT-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
-; DEFAULT-NEXT:    [[IND_END:%.*]] = trunc i64 [[N_VEC]] to i32
+; DEFAULT-NEXT:    [[TMP14:%.*]] = trunc i64 [[N_VEC]] to i32
 ; DEFAULT-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; DEFAULT:       [[VECTOR_BODY]]:
 ; DEFAULT-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
 ; DEFAULT-NEXT:    [[OFFSET_IDX:%.*]] = trunc i64 [[INDEX]] to i32
 ; DEFAULT-NEXT:    [[TMP15:%.*]] = add i32 [[OFFSET_IDX]], 1
 ; DEFAULT-NEXT:    [[TMP16:%.*]] = trunc i64 [[INDEX]] to i32
-; DEFAULT-NEXT:    [[TMP18:%.*]] = add i32 [[TMP16]], 1
-; DEFAULT-NEXT:    [[TMP19:%.*]] = mul i32 [[MUL]], [[TMP16]]
-; DEFAULT-NEXT:    [[TMP20:%.*]] = mul i32 [[MUL]], [[TMP18]]
+; DEFAULT-NEXT:    [[TMP17:%.*]] = add i32 [[TMP16]], 1
+; DEFAULT-NEXT:    [[TMP18:%.*]] = mul i32 [[MUL]], [[TMP16]]
+; DEFAULT-NEXT:    [[TMP19:%.*]] = mul i32 [[MUL]], [[TMP17]]
+; DEFAULT-NEXT:    [[TMP20:%.*]] = zext i32 [[TMP18]] to i64
 ; DEFAULT-NEXT:    [[TMP21:%.*]] = zext i32 [[TMP19]] to i64
-; DEFAULT-NEXT:    [[TMP22:%.*]] = zext i32 [[TMP20]] to i64
+; DEFAULT-NEXT:    [[TMP22:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP20]]
 ; DEFAULT-NEXT:    [[TMP23:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP21]]
-; DEFAULT-NEXT:    [[TMP24:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP22]]
-; DEFAULT-NEXT:    store i32 [[OFFSET_IDX]], ptr [[TMP23]], align 4
-; DEFAULT-NEXT:    store i32 [[TMP15]], ptr [[TMP24]], align 4
+; DEFAULT-NEXT:    store i32 [[OFFSET_IDX]], ptr [[TMP22]], align 4
+; DEFAULT-NEXT:    store i32 [[TMP15]], ptr [[TMP23]], align 4
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; DEFAULT-NEXT:    [[TMP25:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP25]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
+; DEFAULT-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; DEFAULT-NEXT:    br i1 [[TMP24]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
 ; DEFAULT:       [[SCALAR_PH]]:
 ; DEFAULT-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
-; DEFAULT-NEXT:    [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
+; DEFAULT-NEXT:    [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[TMP14]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
 ; DEFAULT-NEXT:    br label %[[LOOP:.*]]
 ; DEFAULT:       [[LOOP]]:
 ; DEFAULT-NEXT:    [[IV_1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_1_NEXT:%.*]], %[[LOOP]] ]
@@ -437,7 +436,7 @@ define void @trunc_ivs_and_store(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 [[IV_2]], ptr [[GEP]], align 4
 ; DEFAULT-NEXT:    [[IV_1_NEXT]] = add i64 [[IV_1]], 1
 ; DEFAULT-NEXT:    [[EXITCOND_3_NOT:%.*]] = icmp eq i64 [[IV_1]], [[N]]
-; DEFAULT-NEXT:    br i1 [[EXITCOND_3_NOT]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP8:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[EXITCOND_3_NOT]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -592,31 +591,31 @@ define void @ivs_trunc_and_ext(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT:       [[VECTOR_PH]]:
 ; DEFAULT-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 2
 ; DEFAULT-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
-; DEFAULT-NEXT:    [[IND_END:%.*]] = trunc i64 [[N_VEC]] to i32
+; DEFAULT-NEXT:    [[TMP13:%.*]] = trunc i64 [[N_VEC]] to i32
 ; DEFAULT-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; DEFAULT:       [[VECTOR_BODY]]:
 ; DEFAULT-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
 ; DEFAULT-NEXT:    [[OFFSET_IDX:%.*]] = trunc i64 [[INDEX]] to i32
 ; DEFAULT-NEXT:    [[TMP14:%.*]] = add i32 [[OFFSET_IDX]], 1
 ; DEFAULT-NEXT:    [[TMP15:%.*]] = trunc i64 [[INDEX]] to i32
-; DEFAULT-NEXT:    [[TMP17:%.*]] = add i32 [[TMP15]], 1
-; DEFAULT-NEXT:    [[TMP18:%.*]] = mul i32 [[ADD]], [[TMP15]]
-; DEFAULT-NEXT:    [[TMP19:%.*]] = mul i32 [[ADD]], [[TMP17]]
+; DEFAULT-NEXT:    [[TMP16:%.*]] = add i32 [[TMP15]], 1
+; DEFAULT-NEXT:    [[TMP17:%.*]] = mul i32 [[ADD]], [[TMP15]]
+; DEFAULT-NEXT:    [[TMP18:%.*]] = mul i32 [[ADD]], [[TMP16]]
+; DEFAULT-NEXT:    [[TMP19:%.*]] = zext i32 [[TMP17]] to i64
 ; DEFAULT-NEXT:    [[TMP20:%.*]] = zext i32 [[TMP18]] to i64
-; DEFAULT-NEXT:    [[TMP21:%.*]] = zext i32 [[TMP19]] to i64
+; DEFAULT-NEXT:    [[TMP21:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP19]]
 ; DEFAULT-NEXT:    [[TMP22:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP20]]
-; DEFAULT-NEXT:    [[TMP23:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP21]]
-; DEFAULT-NEXT:    store i32 [[OFFSET_IDX]], ptr [[TMP22]], align 4
-; DEFAULT-NEXT:    store i32 [[TMP14]], ptr [[TMP23]], align 4
+; DEFAULT-NEXT:    store i32 [[OFFSET_IDX]], ptr [[TMP21]], align 4
+; DEFAULT-NEXT:    store i32 [[TMP14]], ptr [[TMP22]], align 4
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; DEFAULT-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP24]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
+; DEFAULT-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; DEFAULT-NEXT:    br i1 [[TMP23]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
 ; DEFAULT:       [[SCALAR_PH]]:
 ; DEFAULT-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
-; DEFAULT-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[IND_END]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
+; DEFAULT-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[TMP13]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
 ; DEFAULT-NEXT:    br label %[[LOOP:.*]]
 ; DEFAULT:       [[LOOP]]:
 ; DEFAULT-NEXT:    [[IV_1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_1_NEXT:%.*]], %[[LOOP]] ]
@@ -629,7 +628,7 @@ define void @ivs_trunc_and_ext(i32 %x, ptr %dst, i64 %N) #0 {
 ; DEFAULT-NEXT:    store i32 [[IV_2]], ptr [[GEP]], align 4
 ; DEFAULT-NEXT:    [[IV_1_NEXT]] = add i64 [[IV_1]], 1
 ; DEFAULT-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_1]], [[N]]
-; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP10:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP11:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -766,32 +765,32 @@ define void @exit_cond_zext_iv(ptr %dst, i64 %N) {
 ; DEFAULT:       [[VECTOR_SCEVCHECK]]:
 ; DEFAULT-NEXT:    [[UMAX:%.*]] = call i64 @llvm.umax.i64(i64 [[N]], i64 1)
 ; DEFAULT-NEXT:    [[TMP0:%.*]] = add i64 [[UMAX]], -1
-; DEFAULT-NEXT:    [[TMP2:%.*]] = trunc i64 [[TMP0]] to i32
-; DEFAULT-NEXT:    [[TMP3:%.*]] = add i32 1, [[TMP2]]
-; DEFAULT-NEXT:    [[TMP4:%.*]] = icmp ult i32 [[TMP3]], 1
-; DEFAULT-NEXT:    [[TMP5:%.*]] = icmp ugt i64 [[TMP0]], 4294967295
-; DEFAULT-NEXT:    [[TMP6:%.*]] = or i1 [[TMP4]], [[TMP5]]
-; DEFAULT-NEXT:    br i1 [[TMP6]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
+; DEFAULT-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
+; DEFAULT-NEXT:    [[TMP2:%.*]] = add i32 1, [[TMP1]]
+; DEFAULT-NEXT:    [[TMP3:%.*]] = icmp ult i32 [[TMP2]], 1
+; DEFAULT-NEXT:    [[TMP4:%.*]] = icmp ugt i64 [[TMP0]], 4294967295
+; DEFAULT-NEXT:    [[TMP5:%.*]] = or i1 [[TMP3]], [[TMP4]]
+; DEFAULT-NEXT:    br i1 [[TMP5]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
 ; DEFAULT:       [[VECTOR_PH]]:
 ; DEFAULT-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[UMAX1]], 2
 ; DEFAULT-NEXT:    [[N_VEC:%.*]] = sub i64 [[UMAX1]], [[N_MOD_VF]]
-; DEFAULT-NEXT:    [[IND_END:%.*]] = trunc i64 [[N_VEC]] to i32
+; DEFAULT-NEXT:    [[TMP6:%.*]] = trunc i64 [[N_VEC]] to i32
 ; DEFAULT-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; DEFAULT:       [[VECTOR_BODY]]:
 ; DEFAULT-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; DEFAULT-NEXT:    [[TMP8:%.*]] = add i64 [[INDEX]], 1
-; DEFAULT-NEXT:    [[TMP9:%.*]] = getelementptr { [100 x i32], i32, i32 }, ptr [[DST]], i64 [[INDEX]], i32 2
-; DEFAULT-NEXT:    [[TMP10:%.*]] = getelementptr { [100 x i32], i32, i32 }, ptr [[DST]], i64 [[TMP8]], i32 2
+; DEFAULT-NEXT:    [[TMP7:%.*]] = add i64 [[INDEX]], 1
+; DEFAULT-NEXT:    [[TMP8:%.*]] = getelementptr { [100 x i32], i32, i32 }, ptr [[DST]], i64 [[INDEX]], i32 2
+; DEFAULT-NEXT:    [[TMP9:%.*]] = getelementptr { [100 x i32], i32, i32 }, ptr [[DST]], i64 [[TMP7]], i32 2
+; DEFAULT-NEXT:    store i32 0, ptr [[TMP8]], align 8
 ; DEFAULT-NEXT:    store i32 0, ptr [[TMP9]], align 8
-; DEFAULT-NEXT:    store i32 0, ptr [[TMP10]], align 8
 ; DEFAULT-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; DEFAULT-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; DEFAULT-NEXT:    br i1 [[TMP11]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
+; DEFAULT-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; DEFAULT-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
 ; DEFAULT:       [[MIDDLE_BLOCK]]:
 ; DEFAULT-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[UMAX1]], [[N_VEC]]
 ; DEFAULT-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
 ; DEFAULT:       [[SCALAR_PH]]:
-; DEFAULT-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ [[IND_END]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
+; DEFAULT-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ [[TMP6]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
 ; DEFAULT-NEXT:    [[BC_RESUME_VAL2:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ]
 ; DEFAULT-NEXT:    br label %[[LOOP:.*]]
 ; DEFAULT:       [[LOOP]]:
@@ -802,7 +801,7 @@ define void @exit_cond_zext_iv(ptr %dst, i64 %N) {
 ; DEFAULT-NEXT:    [[IV_1_NEXT]] = add i32 [[IV_1]], 1
 ; DEFAULT-NEXT:    [[IV_EXT]] = zext i32 [[IV_1_NEXT]] to i64
 ; DEFAULT-NEXT:    [[C:%.*]] = icmp ult i64 [[IV_EXT]], [[N]]
-; DEFAULT-NEXT:    br i1 [[C]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP12:![0-9]+]]
+; DEFAULT-NEXT:    br i1 [[C]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP13:![0-9]+]]
 ; DEFAULT:       [[EXIT]]:
 ; DEFAULT-NEXT:    ret void
 ;
@@ -814,45 +813,45 @@ define void @exit_cond_zext_iv(ptr %dst, i64 %N) {
 ; PRED:       [[VECTOR_SCEVCHECK]]:
 ; PRED-NEXT:    [[UMAX:%.*]] = call i64 @llvm.umax.i64(i64 [[N]], i64 1)
 ; PRED-NEXT:    [[TMP0:%.*]] = add i64 [[UMAX]], -1
-; PRED-NEXT:    [[TMP2:%.*]] = trunc i64 [[TMP0]] to i32
-; PRED-NEXT:    [[TMP3:%.*]] = add i32 1, [[TMP2]]
-; PRED-NEXT:    [[TMP4:%.*]] = icmp ult i32 [[TMP3]], 1
-; PRED-NEXT:    [[TMP5:%.*]] = icmp ugt i64 [[TMP0]], 4294967295
-; PRED-NEXT:    [[TMP6:%.*]] = or i1 [[TMP4]], [[TMP5]]
-; PRED-NEXT:    br i1 [[TMP6]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
+; PRED-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
+; PRED-NEXT:    [[TMP2:%.*]] = add i32 1, [[TMP1]]
+; PRED-NEXT:    [[TMP3:%.*]] = icmp ult i32 [[TMP2]], 1
+; PRED-NEXT:    [[TMP4:%.*]] = icmp ugt i64 [[TMP0]], 4294967295
+; PRED-NEXT:    [[TMP5:%.*]] = or i1 [[TMP3]], [[TMP4]]
+; PRED-NEXT:    br i1 [[TMP5]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
 ; PRED:       [[VECTOR_PH]]:
 ; PRED-NEXT:    [[N_RND_UP:%.*]] = add i64 [[UMAX1]], 1
 ; PRED-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], 2
 ; PRED-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
 ; PRED-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[UMAX1]], 1
-; PRED-NEXT:    [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
-; PRED-NEXT:    [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
+; PRED-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
+; PRED-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
 ; PRED-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; PRED:       [[VECTOR_BODY]]:
 ; PRED-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_STORE_CONTINUE5:.*]] ]
-; PRED-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[INDEX]], i64 0
-; PRED-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
-; PRED-NEXT:    [[VEC_IV:%.*]] = add <2 x i64> [[BROADCAST_SPLAT]], <i64 0, i64 1>
-; PRED-NEXT:    [[TMP7:%.*]] = icmp ule <2 x i64> [[VEC_IV]], [[BROADCAST_SPLAT3]]
-; PRED-NEXT:    [[TMP8:%.*]] = extractelement <2 x i1> [[TMP7]], i32 0
-; PRED-NEXT:    br i1 [[TMP8]], label %[[PRED_STORE_IF:.*]], label %[[PRED_STORE_CONTINUE:.*]]
+; PRED-NEXT:    [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[INDEX]], i64 0
+; PRED-NEXT:    [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
+; PRED-NEXT:    [[VEC_IV:%.*]] = add <2 x i64> [[BROADCAST_SPLAT3]], <i64 0, i64 1>
+; PRED-NEXT:    [[TMP6:%.*]] = icmp ule <2 x i64> [[VEC_IV]], [[BROADCAST_SPLAT]]
+; PRED-NEXT:    [[TMP7:%.*]] = extractelement <2 x i1> [[TMP6]], i32 0
+; PRED-NEXT:    br i1 [[TMP7]], label %[[PRED_STORE_IF:.*]], label %[[PRED_STORE_CONTINUE:.*]]
 ; PRED:       [[PRED_STORE_IF]]:
-; PRED-NEXT:    [[TMP9:%.*]] = add i64 [[INDEX]], 0
-; PRED-NEXT:    [[TMP10:%.*]] = getelementptr { [100 x i32], i32, i32 }, ptr [[DST]], i64 [[TMP9]], i32 2
-; PRED-NEXT:    store i32 0, ptr [[TMP10]], align 8
+; PRED-NEXT:    [[TMP8:%.*]] = add i64 [[INDEX]], 0
+; PRED-NEXT:    [[TMP9:%.*]] = getelementptr { [100 x i32], i32, i32 }, ptr [[DST]], i64 [[TMP8]], i32 2
+; PRED-NEXT:    store i32 0, ptr [[TMP9]], align 8
 ; PRED-NEXT:    br label %[[PRED_STORE_CONTINUE]]
 ; PRED:       [[PRED_STORE_CONTINUE]]:
-; PRED-NEXT:    [[TMP11:%.*]] = extractelement <2 x i1> [[TMP7]], i32 1
-; PRED-NEXT:    br i1 [[TMP11]], label %[[PRED_STORE_IF4:.*]], label %[[PRED_STORE_CONTINUE5]]
+; PRED-NEXT:    [[TMP10:%.*]] = extractelement <2 x i1> [[TMP6]], i32 1
+; PRED-NEXT:    br i1 [[TMP10]], label %[[PRED_STORE_IF4:.*]], label %[[PRED_STORE_CONTINUE5]]
 ; PRED:       [[PRED_STORE_IF4]]:
-; PRED-NEXT:    [[TMP12:%.*]] = add i64 [[INDEX]], 1
-; PRED-NEXT:    [[TMP13:%.*]] = getelementptr { [100 x i32], i32, i32 }, ptr [[DST]], i64 [[TMP12]], i32 2
-; PRED-NEXT:    store i32 0, ptr [[TMP13]], align 8
+; PRED-NEXT:    [[TMP11:%.*]] = add i64 [[INDEX]], 1
+; PRED-NEXT:    [[TMP12:%.*]] = getelementptr { [100 x i32], i32, i32 }, ptr [[DST]], i64 [[TMP11]], i32 2
+; PRED-NEXT:    store i32 0, ptr [[TMP12]], align 8
 ; PRED-NEXT:    br label %[[PRED_STORE_CONTINUE5]]
 ; PRED:       [[PRED_STORE_CONTINUE5]]:
 ; PRED-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
-; PRED-NEXT:    [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; PRED-NEXT:    br i1 [[TMP14]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
+; PRED-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; PRED-NEXT:    br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
 ; PRED:       [[MIDDLE_BLOCK]]:
 ; PRED-NEXT:    br label %[[EXIT:.*]]
 ; PRED:       [[SCALAR_PH]]:

>From 954acb8aaa86680a4761d60b890f5cb1982caa4b Mon Sep 17 00:00:00 2001
From: Hassnaa Hamdi <hassnaa.hamdi at arm.com>
Date: Tue, 9 Sep 2025 15:53:20 +0000
Subject: [PATCH 3/3] temporarily strip renameing changes from
 interleaving-reduction test

---
 .../AArch64/interleaving-reduction.ll         | 114 +++++++++---------
 1 file changed, 57 insertions(+), 57 deletions(-)

diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/interleaving-reduction.ll b/llvm/test/Transforms/LoopVectorize/AArch64/interleaving-reduction.ll
index 1acd1897fa1f2..46030c9912a32 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/interleaving-reduction.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/interleaving-reduction.ll
@@ -24,30 +24,30 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-4-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; INTERLEAVE-4:       vector.body:
 ; INTERLEAVE-4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI2:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI3:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI4:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
-; INTERLEAVE-4-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 4
-; INTERLEAVE-4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 8
-; INTERLEAVE-4-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 12
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 1
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD5:%.*]] = load <4 x i32>, ptr [[TMP1]], align 1
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD6:%.*]] = load <4 x i32>, ptr [[TMP2]], align 1
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP3]], align 1
-; INTERLEAVE-4-NEXT:    [[TMP4]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
-; INTERLEAVE-4-NEXT:    [[TMP5]] = add <4 x i32> [[VEC_PHI2]], [[WIDE_LOAD5]]
-; INTERLEAVE-4-NEXT:    [[TMP6]] = add <4 x i32> [[VEC_PHI3]], [[WIDE_LOAD6]]
-; INTERLEAVE-4-NEXT:    [[TMP7]] = add <4 x i32> [[VEC_PHI4]], [[WIDE_LOAD7]]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP12:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI2:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI3:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI4:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
+; INTERLEAVE-4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 4
+; INTERLEAVE-4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 8
+; INTERLEAVE-4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 12
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP4]], align 1
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD5:%.*]] = load <4 x i32>, ptr [[TMP9]], align 1
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD6:%.*]] = load <4 x i32>, ptr [[TMP10]], align 1
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP11]], align 1
+; INTERLEAVE-4-NEXT:    [[TMP12]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
+; INTERLEAVE-4-NEXT:    [[TMP13]] = add <4 x i32> [[VEC_PHI2]], [[WIDE_LOAD5]]
+; INTERLEAVE-4-NEXT:    [[TMP14]] = add <4 x i32> [[VEC_PHI3]], [[WIDE_LOAD6]]
+; INTERLEAVE-4-NEXT:    [[TMP15]] = add <4 x i32> [[VEC_PHI4]], [[WIDE_LOAD7]]
 ; INTERLEAVE-4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
-; INTERLEAVE-4-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; INTERLEAVE-4-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; INTERLEAVE-4-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; INTERLEAVE-4-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; INTERLEAVE-4:       middle.block:
-; INTERLEAVE-4-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP5]], [[TMP4]]
-; INTERLEAVE-4-NEXT:    [[BIN_RDX8:%.*]] = add <4 x i32> [[TMP6]], [[BIN_RDX]]
-; INTERLEAVE-4-NEXT:    [[BIN_RDX9:%.*]] = add <4 x i32> [[TMP7]], [[BIN_RDX8]]
-; INTERLEAVE-4-NEXT:    [[TMP9:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX9]])
+; INTERLEAVE-4-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP13]], [[TMP12]]
+; INTERLEAVE-4-NEXT:    [[BIN_RDX8:%.*]] = add <4 x i32> [[TMP14]], [[BIN_RDX]]
+; INTERLEAVE-4-NEXT:    [[BIN_RDX9:%.*]] = add <4 x i32> [[TMP15]], [[BIN_RDX8]]
+; INTERLEAVE-4-NEXT:    [[TMP17:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX9]])
 ; INTERLEAVE-4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
 ; INTERLEAVE-4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
 ; INTERLEAVE-4:       vec.epilog.iter.check:
@@ -56,27 +56,27 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-4-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
 ; INTERLEAVE-4:       vec.epilog.ph:
 ; INTERLEAVE-4-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
-; INTERLEAVE-4-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP9]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; INTERLEAVE-4-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP17]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
 ; INTERLEAVE-4-NEXT:    [[N_MOD_VF10:%.*]] = urem i64 [[N]], 4
 ; INTERLEAVE-4-NEXT:    [[N_VEC11:%.*]] = sub i64 [[N]], [[N_MOD_VF10]]
-; INTERLEAVE-4-NEXT:    [[TMP10:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[BC_MERGE_RDX]], i32 0
+; INTERLEAVE-4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[BC_MERGE_RDX]], i32 0
 ; INTERLEAVE-4-NEXT:    br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
 ; INTERLEAVE-4:       vec.epilog.vector.body:
 ; INTERLEAVE-4-NEXT:    [[INDEX12:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT15:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[VEC_PHI13:%.*]] = phi <4 x i32> [ [[TMP10]], [[VEC_EPILOG_PH]] ], [ [[TMP12:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; INTERLEAVE-4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[INDEX12]]
-; INTERLEAVE-4-NEXT:    [[WIDE_LOAD14:%.*]] = load <4 x i32>, ptr [[TMP11]], align 1
-; INTERLEAVE-4-NEXT:    [[TMP12]] = add <4 x i32> [[VEC_PHI13]], [[WIDE_LOAD14]]
+; INTERLEAVE-4-NEXT:    [[VEC_PHI13:%.*]] = phi <4 x i32> [ [[TMP18]], [[VEC_EPILOG_PH]] ], [ [[TMP22:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; INTERLEAVE-4-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[INDEX12]]
+; INTERLEAVE-4-NEXT:    [[WIDE_LOAD14:%.*]] = load <4 x i32>, ptr [[TMP20]], align 1
+; INTERLEAVE-4-NEXT:    [[TMP22]] = add <4 x i32> [[VEC_PHI13]], [[WIDE_LOAD14]]
 ; INTERLEAVE-4-NEXT:    [[INDEX_NEXT15]] = add nuw i64 [[INDEX12]], 4
-; INTERLEAVE-4-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC11]]
-; INTERLEAVE-4-NEXT:    br i1 [[TMP13]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; INTERLEAVE-4-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC11]]
+; INTERLEAVE-4-NEXT:    br i1 [[TMP23]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
 ; INTERLEAVE-4:       vec.epilog.middle.block:
-; INTERLEAVE-4-NEXT:    [[TMP14:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP12]])
+; INTERLEAVE-4-NEXT:    [[TMP24:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP22]])
 ; INTERLEAVE-4-NEXT:    [[CMP_N16:%.*]] = icmp eq i64 [[N]], [[N_VEC11]]
 ; INTERLEAVE-4-NEXT:    br i1 [[CMP_N16]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
 ; INTERLEAVE-4:       vec.epilog.scalar.ph:
 ; INTERLEAVE-4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC11]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
-; INTERLEAVE-4-NEXT:    [[BC_MERGE_RDX17:%.*]] = phi i32 [ [[TMP14]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[TMP9]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK]] ]
+; INTERLEAVE-4-NEXT:    [[BC_MERGE_RDX17:%.*]] = phi i32 [ [[TMP24]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[TMP17]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK]] ]
 ; INTERLEAVE-4-NEXT:    br label [[LOOP:%.*]]
 ; INTERLEAVE-4:       loop:
 ; INTERLEAVE-4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
@@ -88,7 +88,7 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-4-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
 ; INTERLEAVE-4-NEXT:    br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP4:![0-9]+]]
 ; INTERLEAVE-4:       exit:
-; INTERLEAVE-4-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP9]], [[MIDDLE_BLOCK]] ], [ [[TMP14]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; INTERLEAVE-4-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP17]], [[MIDDLE_BLOCK]] ], [ [[TMP24]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
 ; INTERLEAVE-4-NEXT:    ret i32 [[RED_NEXT_LCSSA]]
 ;
 ; INTERLEAVE-2-LABEL: @interleave_integer_reduction(
@@ -101,25 +101,25 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-2-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; INTERLEAVE-2:       vector.body:
 ; INTERLEAVE-2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-2-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-2-NEXT:    [[VEC_PHI1:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-2-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
-; INTERLEAVE-2-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i32 4
-; INTERLEAVE-2-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 1
-; INTERLEAVE-2-NEXT:    [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP1]], align 1
-; INTERLEAVE-2-NEXT:    [[TMP2]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
-; INTERLEAVE-2-NEXT:    [[TMP3]] = add <4 x i32> [[VEC_PHI1]], [[WIDE_LOAD2]]
+; INTERLEAVE-2-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-2-NEXT:    [[VEC_PHI1:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
+; INTERLEAVE-2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
+; INTERLEAVE-2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 4
+; INTERLEAVE-2-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 1
+; INTERLEAVE-2-NEXT:    [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP5]], align 1
+; INTERLEAVE-2-NEXT:    [[TMP6]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
+; INTERLEAVE-2-NEXT:    [[TMP7]] = add <4 x i32> [[VEC_PHI1]], [[WIDE_LOAD2]]
 ; INTERLEAVE-2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; INTERLEAVE-2-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; INTERLEAVE-2-NEXT:    br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; INTERLEAVE-2-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; INTERLEAVE-2-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; INTERLEAVE-2:       middle.block:
-; INTERLEAVE-2-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP3]], [[TMP2]]
-; INTERLEAVE-2-NEXT:    [[TMP5:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX]])
+; INTERLEAVE-2-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP7]], [[TMP6]]
+; INTERLEAVE-2-NEXT:    [[TMP9:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX]])
 ; INTERLEAVE-2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
 ; INTERLEAVE-2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; INTERLEAVE-2:       scalar.ph:
 ; INTERLEAVE-2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; INTERLEAVE-2-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP5]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; INTERLEAVE-2-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP9]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
 ; INTERLEAVE-2-NEXT:    br label [[LOOP:%.*]]
 ; INTERLEAVE-2:       loop:
 ; INTERLEAVE-2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
@@ -131,7 +131,7 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-2-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
 ; INTERLEAVE-2-NEXT:    br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
 ; INTERLEAVE-2:       exit:
-; INTERLEAVE-2-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP5]], [[MIDDLE_BLOCK]] ]
+; INTERLEAVE-2-NEXT:    [[RED_NEXT_LCSSA:%.*]] = phi i32 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP9]], [[MIDDLE_BLOCK]] ]
 ; INTERLEAVE-2-NEXT:    ret i32 [[RED_NEXT_LCSSA]]
 ;
 ; INTERLEAVE-4-VLA-LABEL: @interleave_integer_reduction(
@@ -166,21 +166,21 @@ define i32 @interleave_integer_reduction(ptr %src, i64 %N) {
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP14:%.*]] = mul nuw i64 [[TMP13]], 12
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i64 [[TMP14]]
 ; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP6]], align 1
-; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD5:%.*]] = load <vscale x 4 x i32>, ptr [[TMP9]], align 1
-; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD6:%.*]] = load <vscale x 4 x i32>, ptr [[TMP12]], align 1
-; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD7:%.*]] = load <vscale x 4 x i32>, ptr [[TMP15]], align 1
+; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD4:%.*]] = load <vscale x 4 x i32>, ptr [[TMP9]], align 1
+; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD5:%.*]] = load <vscale x 4 x i32>, ptr [[TMP12]], align 1
+; INTERLEAVE-4-VLA-NEXT:    [[WIDE_LOAD6:%.*]] = load <vscale x 4 x i32>, ptr [[TMP15]], align 1
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP16]] = add <vscale x 4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
-; INTERLEAVE-4-VLA-NEXT:    [[TMP17]] = add <vscale x 4 x i32> [[VEC_PHI2]], [[WIDE_LOAD5]]
-; INTERLEAVE-4-VLA-NEXT:    [[TMP18]] = add <vscale x 4 x i32> [[VEC_PHI3]], [[WIDE_LOAD6]]
-; INTERLEAVE-4-VLA-NEXT:    [[TMP19]] = add <vscale x 4 x i32> [[VEC_PHI4]], [[WIDE_LOAD7]]
-; INTERLEAVE-4-VLA-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP17]] = add <vscale x 4 x i32> [[VEC_PHI1]], [[WIDE_LOAD4]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP18]] = add <vscale x 4 x i32> [[VEC_PHI2]], [[WIDE_LOAD5]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP19]] = add <vscale x 4 x i32> [[VEC_PHI3]], [[WIDE_LOAD6]]
+; INTERLEAVE-4-VLA-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
 ; INTERLEAVE-4-VLA-NEXT:    [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; INTERLEAVE-4-VLA-NEXT:    br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; INTERLEAVE-4-VLA:       middle.block:
 ; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX:%.*]] = add <vscale x 4 x i32> [[TMP17]], [[TMP16]]
-; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX8:%.*]] = add <vscale x 4 x i32> [[TMP18]], [[BIN_RDX]]
-; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX9:%.*]] = add <vscale x 4 x i32> [[TMP19]], [[BIN_RDX8]]
-; INTERLEAVE-4-VLA-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[BIN_RDX9]])
+; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX7:%.*]] = add <vscale x 4 x i32> [[TMP18]], [[BIN_RDX]]
+; INTERLEAVE-4-VLA-NEXT:    [[BIN_RDX8:%.*]] = add <vscale x 4 x i32> [[TMP19]], [[BIN_RDX7]]
+; INTERLEAVE-4-VLA-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[BIN_RDX8]])
 ; INTERLEAVE-4-VLA-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
 ; INTERLEAVE-4-VLA-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
 ; INTERLEAVE-4-VLA:       vec.epilog.iter.check:



More information about the llvm-commits mailing list