[llvm] be25f52 - [LV] Autogen several tests for ease of update in upcoming change
Philip Reames via llvm-commits
llvm-commits at lists.llvm.org
Wed Jul 20 07:18:15 PDT 2022
Author: Philip Reames
Date: 2022-07-20T07:17:51-07:00
New Revision: be25f52fec1a6e7a659c47799706759ce35d39dd
URL: https://github.com/llvm/llvm-project/commit/be25f52fec1a6e7a659c47799706759ce35d39dd
DIFF: https://github.com/llvm/llvm-project/commit/be25f52fec1a6e7a659c47799706759ce35d39dd.diff
LOG: [LV] Autogen several tests for ease of update in upcoming change
Added:
Modified:
llvm/test/Transforms/LoopVectorize/X86/x86-predication.ll
llvm/test/Transforms/LoopVectorize/induction.ll
llvm/test/Transforms/LoopVectorize/pr44488-predication.ll
llvm/test/Transforms/LoopVectorize/reduction-small-size.ll
Removed:
################################################################################
diff --git a/llvm/test/Transforms/LoopVectorize/X86/x86-predication.ll b/llvm/test/Transforms/LoopVectorize/X86/x86-predication.ll
index 0ba9b207c1f66..514088c26d460 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/x86-predication.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/x86-predication.ll
@@ -1,105 +1,493 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -mattr=avx -force-vector-width=2 -force-vector-interleave=1 -loop-vectorize -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S | FileCheck %s
; RUN: opt -mcpu=skylake-avx512 -S -force-vector-width=8 -force-vector-interleave=1 -loop-vectorize < %s | FileCheck %s --check-prefix=SINK-GATHER
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
target triple = "x86_64-apple-macosx10.8.0"
-; CHECK-LABEL: predicated_sdiv_masked_load
-;
; This test ensures that we don't scalarize the predicated load. Since the load
; can be vectorized with predication, scalarizing it would cause its pointer
; operand to become non-uniform.
;
-; CHECK: vector.body:
-; CHECK: %wide.masked.load = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32
-; CHECK: br i1 {{.*}}, label %[[IF0:.+]], label %[[CONT0:.+]]
-; CHECK: [[IF0]]:
-; CHECK: %[[T0:.+]] = extractelement <2 x i32> %wide.masked.load, i32 0
-; CHECK: %[[T1:.+]] = sdiv i32 %[[T0]], %x
-; CHECK: %[[T2:.+]] = insertelement <2 x i32> poison, i32 %[[T1]], i32 0
-; CHECK: br label %[[CONT0]]
-; CHECK: [[CONT0]]:
-; CHECK: %[[T3:.+]] = phi <2 x i32> [ poison, %vector.body ], [ %[[T2]], %[[IF0]] ]
-; CHECK: br i1 {{.*}}, label %[[IF1:.+]], label %[[CONT1:.+]]
-; CHECK: [[IF1]]:
-; CHECK: %[[T4:.+]] = extractelement <2 x i32> %wide.masked.load, i32 1
-; CHECK: %[[T5:.+]] = sdiv i32 %[[T4]], %x
-; CHECK: %[[T6:.+]] = insertelement <2 x i32> %[[T3]], i32 %[[T5]], i32 1
-; CHECK: br label %[[CONT1]]
-; CHECK: [[CONT1]]:
-; CHECK: phi <2 x i32> [ %[[T3]], %[[CONT0]] ], [ %[[T6]], %[[IF1]] ]
-; CHECK: br i1 {{.*}}, label %middle.block, label %vector.body
-
define i32 @predicated_sdiv_masked_load(i32* %a, i32* %b, i32 %x, i1 %c) {
+; CHECK-LABEL: @predicated_sdiv_masked_load(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i1> poison, i1 [[C:%.*]], i32 0
+; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i1> [[BROADCAST_SPLATINSERT]], <2 x i1> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_SDIV_CONTINUE2:%.*]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x i32> [ zeroinitializer, [[ENTRY]] ], [ [[TMP19:%.*]], [[PRED_SDIV_CONTINUE2]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, i32* [[TMP1]], i32 0
+; CHECK-NEXT: [[TMP3:%.*]] = bitcast i32* [[TMP2]] to <2 x i32>*
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, <2 x i32>* [[TMP3]], align 4
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i32, i32* [[B:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, i32* [[TMP4]], i32 0
+; CHECK-NEXT: [[TMP6:%.*]] = bitcast i32* [[TMP5]] to <2 x i32>*
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* [[TMP6]], i32 4, <2 x i1> [[BROADCAST_SPLAT]], <2 x i32> poison)
+; CHECK-NEXT: [[TMP7:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 0
+; CHECK-NEXT: br i1 [[TMP7]], label [[PRED_SDIV_IF:%.*]], label [[PRED_SDIV_CONTINUE:%.*]]
+; CHECK: pred.sdiv.if:
+; CHECK-NEXT: [[TMP8:%.*]] = extractelement <2 x i32> [[WIDE_MASKED_LOAD]], i32 0
+; CHECK-NEXT: [[TMP9:%.*]] = sdiv i32 [[TMP8]], [[X:%.*]]
+; CHECK-NEXT: [[TMP10:%.*]] = insertelement <2 x i32> poison, i32 [[TMP9]], i32 0
+; CHECK-NEXT: br label [[PRED_SDIV_CONTINUE]]
+; CHECK: pred.sdiv.continue:
+; CHECK-NEXT: [[TMP11:%.*]] = phi <2 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP10]], [[PRED_SDIV_IF]] ]
+; CHECK-NEXT: [[TMP12:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 1
+; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_SDIV_IF1:%.*]], label [[PRED_SDIV_CONTINUE2]]
+; CHECK: pred.sdiv.if1:
+; CHECK-NEXT: [[TMP13:%.*]] = extractelement <2 x i32> [[WIDE_MASKED_LOAD]], i32 1
+; CHECK-NEXT: [[TMP14:%.*]] = sdiv i32 [[TMP13]], [[X]]
+; CHECK-NEXT: [[TMP15:%.*]] = insertelement <2 x i32> [[TMP11]], i32 [[TMP14]], i32 1
+; CHECK-NEXT: br label [[PRED_SDIV_CONTINUE2]]
+; CHECK: pred.sdiv.continue2:
+; CHECK-NEXT: [[TMP16:%.*]] = phi <2 x i32> [ [[TMP11]], [[PRED_SDIV_CONTINUE]] ], [ [[TMP15]], [[PRED_SDIV_IF1]] ]
+; CHECK-NEXT: [[TMP17:%.*]] = add nsw <2 x i32> [[TMP16]], [[WIDE_LOAD]]
+; CHECK-NEXT: [[TMP18:%.*]] = xor <2 x i1> [[BROADCAST_SPLAT]], <i1 true, i1 true>
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[BROADCAST_SPLAT]], <2 x i32> [[TMP17]], <2 x i32> [[WIDE_LOAD]]
+; CHECK-NEXT: [[TMP19]] = add <2 x i32> [[VEC_PHI]], [[PREDPHI]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT]], 10000
+; CHECK-NEXT: br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: middle.block:
+; CHECK-NEXT: [[TMP21:%.*]] = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> [[TMP19]])
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 10000, 10000
+; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ], [ 10000, [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: [[R:%.*]] = phi i32 [ [[T7:%.*]], [[FOR_INC]] ], [ [[TMP21]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[I]]
+; CHECK-NEXT: [[T1:%.*]] = load i32, i32* [[T0]], align 4
+; CHECK-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[FOR_INC]]
+; CHECK: if.then:
+; CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[I]]
+; CHECK-NEXT: [[T3:%.*]] = load i32, i32* [[T2]], align 4
+; CHECK-NEXT: [[T4:%.*]] = sdiv i32 [[T3]], [[X]]
+; CHECK-NEXT: [[T5:%.*]] = add nsw i32 [[T4]], [[T1]]
+; CHECK-NEXT: br label [[FOR_INC]]
+; CHECK: for.inc:
+; CHECK-NEXT: [[T6:%.*]] = phi i32 [ [[T1]], [[FOR_BODY]] ], [ [[T5]], [[IF_THEN]] ]
+; CHECK-NEXT: [[T7]] = add i32 [[R]], [[T6]]
+; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
+; CHECK-NEXT: [[COND:%.*]] = icmp eq i64 [[I_NEXT]], 10000
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
+; CHECK: for.end:
+; CHECK-NEXT: [[T8:%.*]] = phi i32 [ [[T7]], [[FOR_INC]] ], [ [[TMP21]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: ret i32 [[T8]]
+;
+; SINK-GATHER-LABEL: @predicated_sdiv_masked_load(
+; SINK-GATHER-NEXT: entry:
+; SINK-GATHER-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; SINK-GATHER: vector.ph:
+; SINK-GATHER-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <8 x i1> poison, i1 [[C:%.*]], i32 0
+; SINK-GATHER-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <8 x i1> [[BROADCAST_SPLATINSERT]], <8 x i1> poison, <8 x i32> zeroinitializer
+; SINK-GATHER-NEXT: br label [[VECTOR_BODY:%.*]]
+; SINK-GATHER: vector.body:
+; SINK-GATHER-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_SDIV_CONTINUE14:%.*]] ]
+; SINK-GATHER-NEXT: [[VEC_PHI:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP49:%.*]], [[PRED_SDIV_CONTINUE14]] ]
+; SINK-GATHER-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; SINK-GATHER-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[TMP0]]
+; SINK-GATHER-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, i32* [[TMP1]], i32 0
+; SINK-GATHER-NEXT: [[TMP3:%.*]] = bitcast i32* [[TMP2]] to <8 x i32>*
+; SINK-GATHER-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i32>, <8 x i32>* [[TMP3]], align 4
+; SINK-GATHER-NEXT: [[TMP4:%.*]] = getelementptr i32, i32* [[B:%.*]], i64 [[TMP0]]
+; SINK-GATHER-NEXT: [[TMP5:%.*]] = getelementptr i32, i32* [[TMP4]], i32 0
+; SINK-GATHER-NEXT: [[TMP6:%.*]] = bitcast i32* [[TMP5]] to <8 x i32>*
+; SINK-GATHER-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* [[TMP6]], i32 4, <8 x i1> [[BROADCAST_SPLAT]], <8 x i32> poison)
+; SINK-GATHER-NEXT: [[TMP7:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 0
+; SINK-GATHER-NEXT: br i1 [[TMP7]], label [[PRED_SDIV_IF:%.*]], label [[PRED_SDIV_CONTINUE:%.*]]
+; SINK-GATHER: pred.sdiv.if:
+; SINK-GATHER-NEXT: [[TMP8:%.*]] = extractelement <8 x i32> [[WIDE_MASKED_LOAD]], i32 0
+; SINK-GATHER-NEXT: [[TMP9:%.*]] = sdiv i32 [[TMP8]], [[X:%.*]]
+; SINK-GATHER-NEXT: [[TMP10:%.*]] = insertelement <8 x i32> poison, i32 [[TMP9]], i32 0
+; SINK-GATHER-NEXT: br label [[PRED_SDIV_CONTINUE]]
+; SINK-GATHER: pred.sdiv.continue:
+; SINK-GATHER-NEXT: [[TMP11:%.*]] = phi <8 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP10]], [[PRED_SDIV_IF]] ]
+; SINK-GATHER-NEXT: [[TMP12:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 1
+; SINK-GATHER-NEXT: br i1 [[TMP12]], label [[PRED_SDIV_IF1:%.*]], label [[PRED_SDIV_CONTINUE2:%.*]]
+; SINK-GATHER: pred.sdiv.if1:
+; SINK-GATHER-NEXT: [[TMP13:%.*]] = extractelement <8 x i32> [[WIDE_MASKED_LOAD]], i32 1
+; SINK-GATHER-NEXT: [[TMP14:%.*]] = sdiv i32 [[TMP13]], [[X]]
+; SINK-GATHER-NEXT: [[TMP15:%.*]] = insertelement <8 x i32> [[TMP11]], i32 [[TMP14]], i32 1
+; SINK-GATHER-NEXT: br label [[PRED_SDIV_CONTINUE2]]
+; SINK-GATHER: pred.sdiv.continue2:
+; SINK-GATHER-NEXT: [[TMP16:%.*]] = phi <8 x i32> [ [[TMP11]], [[PRED_SDIV_CONTINUE]] ], [ [[TMP15]], [[PRED_SDIV_IF1]] ]
+; SINK-GATHER-NEXT: [[TMP17:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 2
+; SINK-GATHER-NEXT: br i1 [[TMP17]], label [[PRED_SDIV_IF3:%.*]], label [[PRED_SDIV_CONTINUE4:%.*]]
+; SINK-GATHER: pred.sdiv.if3:
+; SINK-GATHER-NEXT: [[TMP18:%.*]] = extractelement <8 x i32> [[WIDE_MASKED_LOAD]], i32 2
+; SINK-GATHER-NEXT: [[TMP19:%.*]] = sdiv i32 [[TMP18]], [[X]]
+; SINK-GATHER-NEXT: [[TMP20:%.*]] = insertelement <8 x i32> [[TMP16]], i32 [[TMP19]], i32 2
+; SINK-GATHER-NEXT: br label [[PRED_SDIV_CONTINUE4]]
+; SINK-GATHER: pred.sdiv.continue4:
+; SINK-GATHER-NEXT: [[TMP21:%.*]] = phi <8 x i32> [ [[TMP16]], [[PRED_SDIV_CONTINUE2]] ], [ [[TMP20]], [[PRED_SDIV_IF3]] ]
+; SINK-GATHER-NEXT: [[TMP22:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 3
+; SINK-GATHER-NEXT: br i1 [[TMP22]], label [[PRED_SDIV_IF5:%.*]], label [[PRED_SDIV_CONTINUE6:%.*]]
+; SINK-GATHER: pred.sdiv.if5:
+; SINK-GATHER-NEXT: [[TMP23:%.*]] = extractelement <8 x i32> [[WIDE_MASKED_LOAD]], i32 3
+; SINK-GATHER-NEXT: [[TMP24:%.*]] = sdiv i32 [[TMP23]], [[X]]
+; SINK-GATHER-NEXT: [[TMP25:%.*]] = insertelement <8 x i32> [[TMP21]], i32 [[TMP24]], i32 3
+; SINK-GATHER-NEXT: br label [[PRED_SDIV_CONTINUE6]]
+; SINK-GATHER: pred.sdiv.continue6:
+; SINK-GATHER-NEXT: [[TMP26:%.*]] = phi <8 x i32> [ [[TMP21]], [[PRED_SDIV_CONTINUE4]] ], [ [[TMP25]], [[PRED_SDIV_IF5]] ]
+; SINK-GATHER-NEXT: [[TMP27:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 4
+; SINK-GATHER-NEXT: br i1 [[TMP27]], label [[PRED_SDIV_IF7:%.*]], label [[PRED_SDIV_CONTINUE8:%.*]]
+; SINK-GATHER: pred.sdiv.if7:
+; SINK-GATHER-NEXT: [[TMP28:%.*]] = extractelement <8 x i32> [[WIDE_MASKED_LOAD]], i32 4
+; SINK-GATHER-NEXT: [[TMP29:%.*]] = sdiv i32 [[TMP28]], [[X]]
+; SINK-GATHER-NEXT: [[TMP30:%.*]] = insertelement <8 x i32> [[TMP26]], i32 [[TMP29]], i32 4
+; SINK-GATHER-NEXT: br label [[PRED_SDIV_CONTINUE8]]
+; SINK-GATHER: pred.sdiv.continue8:
+; SINK-GATHER-NEXT: [[TMP31:%.*]] = phi <8 x i32> [ [[TMP26]], [[PRED_SDIV_CONTINUE6]] ], [ [[TMP30]], [[PRED_SDIV_IF7]] ]
+; SINK-GATHER-NEXT: [[TMP32:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 5
+; SINK-GATHER-NEXT: br i1 [[TMP32]], label [[PRED_SDIV_IF9:%.*]], label [[PRED_SDIV_CONTINUE10:%.*]]
+; SINK-GATHER: pred.sdiv.if9:
+; SINK-GATHER-NEXT: [[TMP33:%.*]] = extractelement <8 x i32> [[WIDE_MASKED_LOAD]], i32 5
+; SINK-GATHER-NEXT: [[TMP34:%.*]] = sdiv i32 [[TMP33]], [[X]]
+; SINK-GATHER-NEXT: [[TMP35:%.*]] = insertelement <8 x i32> [[TMP31]], i32 [[TMP34]], i32 5
+; SINK-GATHER-NEXT: br label [[PRED_SDIV_CONTINUE10]]
+; SINK-GATHER: pred.sdiv.continue10:
+; SINK-GATHER-NEXT: [[TMP36:%.*]] = phi <8 x i32> [ [[TMP31]], [[PRED_SDIV_CONTINUE8]] ], [ [[TMP35]], [[PRED_SDIV_IF9]] ]
+; SINK-GATHER-NEXT: [[TMP37:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 6
+; SINK-GATHER-NEXT: br i1 [[TMP37]], label [[PRED_SDIV_IF11:%.*]], label [[PRED_SDIV_CONTINUE12:%.*]]
+; SINK-GATHER: pred.sdiv.if11:
+; SINK-GATHER-NEXT: [[TMP38:%.*]] = extractelement <8 x i32> [[WIDE_MASKED_LOAD]], i32 6
+; SINK-GATHER-NEXT: [[TMP39:%.*]] = sdiv i32 [[TMP38]], [[X]]
+; SINK-GATHER-NEXT: [[TMP40:%.*]] = insertelement <8 x i32> [[TMP36]], i32 [[TMP39]], i32 6
+; SINK-GATHER-NEXT: br label [[PRED_SDIV_CONTINUE12]]
+; SINK-GATHER: pred.sdiv.continue12:
+; SINK-GATHER-NEXT: [[TMP41:%.*]] = phi <8 x i32> [ [[TMP36]], [[PRED_SDIV_CONTINUE10]] ], [ [[TMP40]], [[PRED_SDIV_IF11]] ]
+; SINK-GATHER-NEXT: [[TMP42:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 7
+; SINK-GATHER-NEXT: br i1 [[TMP42]], label [[PRED_SDIV_IF13:%.*]], label [[PRED_SDIV_CONTINUE14]]
+; SINK-GATHER: pred.sdiv.if13:
+; SINK-GATHER-NEXT: [[TMP43:%.*]] = extractelement <8 x i32> [[WIDE_MASKED_LOAD]], i32 7
+; SINK-GATHER-NEXT: [[TMP44:%.*]] = sdiv i32 [[TMP43]], [[X]]
+; SINK-GATHER-NEXT: [[TMP45:%.*]] = insertelement <8 x i32> [[TMP41]], i32 [[TMP44]], i32 7
+; SINK-GATHER-NEXT: br label [[PRED_SDIV_CONTINUE14]]
+; SINK-GATHER: pred.sdiv.continue14:
+; SINK-GATHER-NEXT: [[TMP46:%.*]] = phi <8 x i32> [ [[TMP41]], [[PRED_SDIV_CONTINUE12]] ], [ [[TMP45]], [[PRED_SDIV_IF13]] ]
+; SINK-GATHER-NEXT: [[TMP47:%.*]] = add nsw <8 x i32> [[TMP46]], [[WIDE_LOAD]]
+; SINK-GATHER-NEXT: [[TMP48:%.*]] = xor <8 x i1> [[BROADCAST_SPLAT]], <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>
+; SINK-GATHER-NEXT: [[PREDPHI:%.*]] = select <8 x i1> [[BROADCAST_SPLAT]], <8 x i32> [[TMP47]], <8 x i32> [[WIDE_LOAD]]
+; SINK-GATHER-NEXT: [[TMP49]] = add <8 x i32> [[VEC_PHI]], [[PREDPHI]]
+; SINK-GATHER-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
+; SINK-GATHER-NEXT: [[TMP50:%.*]] = icmp eq i64 [[INDEX_NEXT]], 10000
+; SINK-GATHER-NEXT: br i1 [[TMP50]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; SINK-GATHER: middle.block:
+; SINK-GATHER-NEXT: [[TMP51:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP49]])
+; SINK-GATHER-NEXT: [[CMP_N:%.*]] = icmp eq i64 10000, 10000
+; SINK-GATHER-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
+; SINK-GATHER: scalar.ph:
+; SINK-GATHER-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 10000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; SINK-GATHER-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[TMP51]], [[MIDDLE_BLOCK]] ]
+; SINK-GATHER-NEXT: br label [[FOR_BODY:%.*]]
+; SINK-GATHER: for.body:
+; SINK-GATHER-NEXT: [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ]
+; SINK-GATHER-NEXT: [[R:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[T7:%.*]], [[FOR_INC]] ]
+; SINK-GATHER-NEXT: [[T0:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[I]]
+; SINK-GATHER-NEXT: [[T1:%.*]] = load i32, i32* [[T0]], align 4
+; SINK-GATHER-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[FOR_INC]]
+; SINK-GATHER: if.then:
+; SINK-GATHER-NEXT: [[T2:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[I]]
+; SINK-GATHER-NEXT: [[T3:%.*]] = load i32, i32* [[T2]], align 4
+; SINK-GATHER-NEXT: [[T4:%.*]] = sdiv i32 [[T3]], [[X]]
+; SINK-GATHER-NEXT: [[T5:%.*]] = add nsw i32 [[T4]], [[T1]]
+; SINK-GATHER-NEXT: br label [[FOR_INC]]
+; SINK-GATHER: for.inc:
+; SINK-GATHER-NEXT: [[T6:%.*]] = phi i32 [ [[T1]], [[FOR_BODY]] ], [ [[T5]], [[IF_THEN]] ]
+; SINK-GATHER-NEXT: [[T7]] = add i32 [[R]], [[T6]]
+; SINK-GATHER-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
+; SINK-GATHER-NEXT: [[COND:%.*]] = icmp eq i64 [[I_NEXT]], 10000
+; SINK-GATHER-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
+; SINK-GATHER: for.end:
+; SINK-GATHER-NEXT: [[T8:%.*]] = phi i32 [ [[T7]], [[FOR_INC]] ], [ [[TMP51]], [[MIDDLE_BLOCK]] ]
+; SINK-GATHER-NEXT: ret i32 [[T8]]
+;
entry:
br label %for.body
for.body:
%i = phi i64 [ 0, %entry ], [ %i.next, %for.inc ]
- %r = phi i32 [ 0, %entry ], [ %tmp7, %for.inc ]
- %tmp0 = getelementptr inbounds i32, i32* %a, i64 %i
- %tmp1 = load i32, i32* %tmp0, align 4
+ %r = phi i32 [ 0, %entry ], [ %t7, %for.inc ]
+ %t0 = getelementptr inbounds i32, i32* %a, i64 %i
+ %t1 = load i32, i32* %t0, align 4
br i1 %c, label %if.then, label %for.inc
if.then:
- %tmp2 = getelementptr inbounds i32, i32* %b, i64 %i
- %tmp3 = load i32, i32* %tmp2, align 4
- %tmp4 = sdiv i32 %tmp3, %x
- %tmp5 = add nsw i32 %tmp4, %tmp1
+ %t2 = getelementptr inbounds i32, i32* %b, i64 %i
+ %t3 = load i32, i32* %t2, align 4
+ %t4 = sdiv i32 %t3, %x
+ %t5 = add nsw i32 %t4, %t1
br label %for.inc
for.inc:
- %tmp6 = phi i32 [ %tmp1, %for.body ], [ %tmp5, %if.then]
- %tmp7 = add i32 %r, %tmp6
+ %t6 = phi i32 [ %t1, %for.body ], [ %t5, %if.then]
+ %t7 = add i32 %r, %t6
%i.next = add nuw nsw i64 %i, 1
%cond = icmp eq i64 %i.next, 10000
br i1 %cond, label %for.end, label %for.body
for.end:
- %tmp8 = phi i32 [ %tmp7, %for.inc ]
- ret i32 %tmp8
+ %t8 = phi i32 [ %t7, %for.inc ]
+ ret i32 %t8
}
; This test ensures that a load, which would have been widened otherwise is
; instead scalarized if Cost-Model so decided as part of its
; sink-scalar-operands optimization for predicated instructions.
-;
-; SINK-GATHER-LABEL: @scalarize_and_sink_gather
-; SINK-GATHER: vector.body:
-; SINK-GATHER-LABEL: pred.udiv.if: ; preds = %vector.body
-; SINK-GATHER-NEXT: [[EXT:%.+]] = extractelement <8 x i64> {{.*}}, i32 0
-; SINK-GATHER-NEXT: [[GEP:%.+]] = getelementptr inbounds i32, i32* %a, i64 [[EXT]]
-; SINK-GATHER-NEXT: [[LV:%.+]] = load i32, i32* [[GEP]], align 4
-; SINK-GATHER-NEXT: [[UDIV:%.+]] = udiv i32 [[LV]], %x
-; SINK-GATHER-NEXT: [[INS:%.+]] = insertelement <8 x i32> poison, i32 [[UDIV]], i32 0
-; SINK-GATHER-NEXT: br label %pred.udiv.continue
-; SINK-GATHER: pred.udiv.continue:
-; SINK-GATHER-NEXT: phi i32 [ poison, %vector.body ], [ [[LV]], %pred.udiv.if ]
-; SINK-GATHER-NEXT: phi <8 x i32> [ poison, %vector.body ], [ [[INS]], %pred.udiv.if ]
define i32 @scalarize_and_sink_gather(i32* %a, i1 %c, i32 %x, i64 %n) {
+; CHECK-LABEL: @scalarize_and_sink_gather(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N:%.*]], i64 1)
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[SMAX]], 2
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK: vector.ph:
+; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[SMAX]], 2
+; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[SMAX]], [[N_MOD_VF]]
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i1> poison, i1 [[C:%.*]], i32 0
+; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i1> [[BROADCAST_SPLATINSERT]], <2 x i1> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT3:%.*]] = insertelement <2 x i32> poison, i32 [[X:%.*]], i32 0
+; CHECK-NEXT: [[BROADCAST_SPLAT4:%.*]] = shufflevector <2 x i32> [[BROADCAST_SPLATINSERT3]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_UDIV_CONTINUE2:%.*]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_UDIV_CONTINUE2]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[PRED_UDIV_CONTINUE2]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = mul <2 x i64> [[VEC_IND]], <i64 777, i64 777>
+; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 0
+; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
+; CHECK: pred.udiv.if:
+; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x i64> [[TMP0]], i32 0
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
+; CHECK-NEXT: [[TMP5:%.*]] = udiv i32 [[TMP4]], [[X]]
+; CHECK-NEXT: [[TMP6:%.*]] = insertelement <2 x i32> poison, i32 [[TMP5]], i32 0
+; CHECK-NEXT: br label [[PRED_UDIV_CONTINUE]]
+; CHECK: pred.udiv.continue:
+; CHECK-NEXT: [[TMP7:%.*]] = phi i32 [ poison, [[VECTOR_BODY]] ], [ [[TMP4]], [[PRED_UDIV_IF]] ]
+; CHECK-NEXT: [[TMP8:%.*]] = phi <2 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP6]], [[PRED_UDIV_IF]] ]
+; CHECK-NEXT: [[TMP9:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 1
+; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_UDIV_IF1:%.*]], label [[PRED_UDIV_CONTINUE2]]
+; CHECK: pred.udiv.if1:
+; CHECK-NEXT: [[TMP10:%.*]] = extractelement <2 x i64> [[TMP0]], i32 1
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
+; CHECK-NEXT: [[TMP13:%.*]] = udiv i32 [[TMP12]], [[X]]
+; CHECK-NEXT: [[TMP14:%.*]] = insertelement <2 x i32> [[TMP8]], i32 [[TMP13]], i32 1
+; CHECK-NEXT: br label [[PRED_UDIV_CONTINUE2]]
+; CHECK: pred.udiv.continue2:
+; CHECK-NEXT: [[TMP15:%.*]] = phi i32 [ poison, [[PRED_UDIV_CONTINUE]] ], [ [[TMP12]], [[PRED_UDIV_IF1]] ]
+; CHECK-NEXT: [[TMP16:%.*]] = phi <2 x i32> [ [[TMP8]], [[PRED_UDIV_CONTINUE]] ], [ [[TMP14]], [[PRED_UDIV_IF1]] ]
+; CHECK-NEXT: [[TMP17:%.*]] = xor <2 x i1> [[BROADCAST_SPLAT]], <i1 true, i1 true>
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[BROADCAST_SPLAT]], <2 x i32> [[TMP16]], <2 x i32> [[BROADCAST_SPLAT4]]
+; CHECK-NEXT: [[TMP18]] = add <2 x i32> [[VEC_PHI]], [[PREDPHI]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; CHECK-NEXT: [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK: middle.block:
+; CHECK-NEXT: [[TMP20:%.*]] = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> [[TMP18]])
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[SMAX]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
+; CHECK: scalar.ph:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[TMP20]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ]
+; CHECK-NEXT: [[R:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[T6:%.*]], [[FOR_INC]] ]
+; CHECK-NEXT: [[I7:%.*]] = mul i64 [[I]], 777
+; CHECK-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[FOR_INC]]
+; CHECK: if.then:
+; CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[I7]]
+; CHECK-NEXT: [[T2:%.*]] = load i32, i32* [[T0]], align 4
+; CHECK-NEXT: [[T4:%.*]] = udiv i32 [[T2]], [[X]]
+; CHECK-NEXT: br label [[FOR_INC]]
+; CHECK: for.inc:
+; CHECK-NEXT: [[T5:%.*]] = phi i32 [ [[X]], [[FOR_BODY]] ], [ [[T4]], [[IF_THEN]] ]
+; CHECK-NEXT: [[T6]] = add i32 [[R]], [[T5]]
+; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
+; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK: for.end:
+; CHECK-NEXT: [[T7:%.*]] = phi i32 [ [[T6]], [[FOR_INC]] ], [ [[TMP20]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: ret i32 [[T7]]
+;
+; SINK-GATHER-LABEL: @scalarize_and_sink_gather(
+; SINK-GATHER-NEXT: entry:
+; SINK-GATHER-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N:%.*]], i64 1)
+; SINK-GATHER-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[SMAX]], 8
+; SINK-GATHER-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; SINK-GATHER: vector.ph:
+; SINK-GATHER-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[SMAX]], 8
+; SINK-GATHER-NEXT: [[N_VEC:%.*]] = sub i64 [[SMAX]], [[N_MOD_VF]]
+; SINK-GATHER-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <8 x i1> poison, i1 [[C:%.*]], i32 0
+; SINK-GATHER-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <8 x i1> [[BROADCAST_SPLATINSERT]], <8 x i1> poison, <8 x i32> zeroinitializer
+; SINK-GATHER-NEXT: [[BROADCAST_SPLATINSERT15:%.*]] = insertelement <8 x i32> poison, i32 [[X:%.*]], i32 0
+; SINK-GATHER-NEXT: [[BROADCAST_SPLAT16:%.*]] = shufflevector <8 x i32> [[BROADCAST_SPLATINSERT15]], <8 x i32> poison, <8 x i32> zeroinitializer
+; SINK-GATHER-NEXT: br label [[VECTOR_BODY:%.*]]
+; SINK-GATHER: vector.body:
+; SINK-GATHER-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_UDIV_CONTINUE14:%.*]] ]
+; SINK-GATHER-NEXT: [[VEC_IND:%.*]] = phi <8 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_UDIV_CONTINUE14]] ]
+; SINK-GATHER-NEXT: [[VEC_PHI:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP66:%.*]], [[PRED_UDIV_CONTINUE14]] ]
+; SINK-GATHER-NEXT: [[TMP0:%.*]] = mul <8 x i64> [[VEC_IND]], <i64 777, i64 777, i64 777, i64 777, i64 777, i64 777, i64 777, i64 777>
+; SINK-GATHER-NEXT: [[TMP1:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 0
+; SINK-GATHER-NEXT: br i1 [[TMP1]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
+; SINK-GATHER: pred.udiv.if:
+; SINK-GATHER-NEXT: [[TMP2:%.*]] = extractelement <8 x i64> [[TMP0]], i32 0
+; SINK-GATHER-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[TMP2]]
+; SINK-GATHER-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
+; SINK-GATHER-NEXT: [[TMP5:%.*]] = udiv i32 [[TMP4]], [[X]]
+; SINK-GATHER-NEXT: [[TMP6:%.*]] = insertelement <8 x i32> poison, i32 [[TMP5]], i32 0
+; SINK-GATHER-NEXT: br label [[PRED_UDIV_CONTINUE]]
+; SINK-GATHER: pred.udiv.continue:
+; SINK-GATHER-NEXT: [[TMP7:%.*]] = phi i32 [ poison, [[VECTOR_BODY]] ], [ [[TMP4]], [[PRED_UDIV_IF]] ]
+; SINK-GATHER-NEXT: [[TMP8:%.*]] = phi <8 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP6]], [[PRED_UDIV_IF]] ]
+; SINK-GATHER-NEXT: [[TMP9:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 1
+; SINK-GATHER-NEXT: br i1 [[TMP9]], label [[PRED_UDIV_IF1:%.*]], label [[PRED_UDIV_CONTINUE2:%.*]]
+; SINK-GATHER: pred.udiv.if1:
+; SINK-GATHER-NEXT: [[TMP10:%.*]] = extractelement <8 x i64> [[TMP0]], i32 1
+; SINK-GATHER-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP10]]
+; SINK-GATHER-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
+; SINK-GATHER-NEXT: [[TMP13:%.*]] = udiv i32 [[TMP12]], [[X]]
+; SINK-GATHER-NEXT: [[TMP14:%.*]] = insertelement <8 x i32> [[TMP8]], i32 [[TMP13]], i32 1
+; SINK-GATHER-NEXT: br label [[PRED_UDIV_CONTINUE2]]
+; SINK-GATHER: pred.udiv.continue2:
+; SINK-GATHER-NEXT: [[TMP15:%.*]] = phi i32 [ poison, [[PRED_UDIV_CONTINUE]] ], [ [[TMP12]], [[PRED_UDIV_IF1]] ]
+; SINK-GATHER-NEXT: [[TMP16:%.*]] = phi <8 x i32> [ [[TMP8]], [[PRED_UDIV_CONTINUE]] ], [ [[TMP14]], [[PRED_UDIV_IF1]] ]
+; SINK-GATHER-NEXT: [[TMP17:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 2
+; SINK-GATHER-NEXT: br i1 [[TMP17]], label [[PRED_UDIV_IF3:%.*]], label [[PRED_UDIV_CONTINUE4:%.*]]
+; SINK-GATHER: pred.udiv.if3:
+; SINK-GATHER-NEXT: [[TMP18:%.*]] = extractelement <8 x i64> [[TMP0]], i32 2
+; SINK-GATHER-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP18]]
+; SINK-GATHER-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4
+; SINK-GATHER-NEXT: [[TMP21:%.*]] = udiv i32 [[TMP20]], [[X]]
+; SINK-GATHER-NEXT: [[TMP22:%.*]] = insertelement <8 x i32> [[TMP16]], i32 [[TMP21]], i32 2
+; SINK-GATHER-NEXT: br label [[PRED_UDIV_CONTINUE4]]
+; SINK-GATHER: pred.udiv.continue4:
+; SINK-GATHER-NEXT: [[TMP23:%.*]] = phi i32 [ poison, [[PRED_UDIV_CONTINUE2]] ], [ [[TMP20]], [[PRED_UDIV_IF3]] ]
+; SINK-GATHER-NEXT: [[TMP24:%.*]] = phi <8 x i32> [ [[TMP16]], [[PRED_UDIV_CONTINUE2]] ], [ [[TMP22]], [[PRED_UDIV_IF3]] ]
+; SINK-GATHER-NEXT: [[TMP25:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 3
+; SINK-GATHER-NEXT: br i1 [[TMP25]], label [[PRED_UDIV_IF5:%.*]], label [[PRED_UDIV_CONTINUE6:%.*]]
+; SINK-GATHER: pred.udiv.if5:
+; SINK-GATHER-NEXT: [[TMP26:%.*]] = extractelement <8 x i64> [[TMP0]], i32 3
+; SINK-GATHER-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP26]]
+; SINK-GATHER-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
+; SINK-GATHER-NEXT: [[TMP29:%.*]] = udiv i32 [[TMP28]], [[X]]
+; SINK-GATHER-NEXT: [[TMP30:%.*]] = insertelement <8 x i32> [[TMP24]], i32 [[TMP29]], i32 3
+; SINK-GATHER-NEXT: br label [[PRED_UDIV_CONTINUE6]]
+; SINK-GATHER: pred.udiv.continue6:
+; SINK-GATHER-NEXT: [[TMP31:%.*]] = phi i32 [ poison, [[PRED_UDIV_CONTINUE4]] ], [ [[TMP28]], [[PRED_UDIV_IF5]] ]
+; SINK-GATHER-NEXT: [[TMP32:%.*]] = phi <8 x i32> [ [[TMP24]], [[PRED_UDIV_CONTINUE4]] ], [ [[TMP30]], [[PRED_UDIV_IF5]] ]
+; SINK-GATHER-NEXT: [[TMP33:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 4
+; SINK-GATHER-NEXT: br i1 [[TMP33]], label [[PRED_UDIV_IF7:%.*]], label [[PRED_UDIV_CONTINUE8:%.*]]
+; SINK-GATHER: pred.udiv.if7:
+; SINK-GATHER-NEXT: [[TMP34:%.*]] = extractelement <8 x i64> [[TMP0]], i32 4
+; SINK-GATHER-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP34]]
+; SINK-GATHER-NEXT: [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4
+; SINK-GATHER-NEXT: [[TMP37:%.*]] = udiv i32 [[TMP36]], [[X]]
+; SINK-GATHER-NEXT: [[TMP38:%.*]] = insertelement <8 x i32> [[TMP32]], i32 [[TMP37]], i32 4
+; SINK-GATHER-NEXT: br label [[PRED_UDIV_CONTINUE8]]
+; SINK-GATHER: pred.udiv.continue8:
+; SINK-GATHER-NEXT: [[TMP39:%.*]] = phi i32 [ poison, [[PRED_UDIV_CONTINUE6]] ], [ [[TMP36]], [[PRED_UDIV_IF7]] ]
+; SINK-GATHER-NEXT: [[TMP40:%.*]] = phi <8 x i32> [ [[TMP32]], [[PRED_UDIV_CONTINUE6]] ], [ [[TMP38]], [[PRED_UDIV_IF7]] ]
+; SINK-GATHER-NEXT: [[TMP41:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 5
+; SINK-GATHER-NEXT: br i1 [[TMP41]], label [[PRED_UDIV_IF9:%.*]], label [[PRED_UDIV_CONTINUE10:%.*]]
+; SINK-GATHER: pred.udiv.if9:
+; SINK-GATHER-NEXT: [[TMP42:%.*]] = extractelement <8 x i64> [[TMP0]], i32 5
+; SINK-GATHER-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP42]]
+; SINK-GATHER-NEXT: [[TMP44:%.*]] = load i32, i32* [[TMP43]], align 4
+; SINK-GATHER-NEXT: [[TMP45:%.*]] = udiv i32 [[TMP44]], [[X]]
+; SINK-GATHER-NEXT: [[TMP46:%.*]] = insertelement <8 x i32> [[TMP40]], i32 [[TMP45]], i32 5
+; SINK-GATHER-NEXT: br label [[PRED_UDIV_CONTINUE10]]
+; SINK-GATHER: pred.udiv.continue10:
+; SINK-GATHER-NEXT: [[TMP47:%.*]] = phi i32 [ poison, [[PRED_UDIV_CONTINUE8]] ], [ [[TMP44]], [[PRED_UDIV_IF9]] ]
+; SINK-GATHER-NEXT: [[TMP48:%.*]] = phi <8 x i32> [ [[TMP40]], [[PRED_UDIV_CONTINUE8]] ], [ [[TMP46]], [[PRED_UDIV_IF9]] ]
+; SINK-GATHER-NEXT: [[TMP49:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 6
+; SINK-GATHER-NEXT: br i1 [[TMP49]], label [[PRED_UDIV_IF11:%.*]], label [[PRED_UDIV_CONTINUE12:%.*]]
+; SINK-GATHER: pred.udiv.if11:
+; SINK-GATHER-NEXT: [[TMP50:%.*]] = extractelement <8 x i64> [[TMP0]], i32 6
+; SINK-GATHER-NEXT: [[TMP51:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP50]]
+; SINK-GATHER-NEXT: [[TMP52:%.*]] = load i32, i32* [[TMP51]], align 4
+; SINK-GATHER-NEXT: [[TMP53:%.*]] = udiv i32 [[TMP52]], [[X]]
+; SINK-GATHER-NEXT: [[TMP54:%.*]] = insertelement <8 x i32> [[TMP48]], i32 [[TMP53]], i32 6
+; SINK-GATHER-NEXT: br label [[PRED_UDIV_CONTINUE12]]
+; SINK-GATHER: pred.udiv.continue12:
+; SINK-GATHER-NEXT: [[TMP55:%.*]] = phi i32 [ poison, [[PRED_UDIV_CONTINUE10]] ], [ [[TMP52]], [[PRED_UDIV_IF11]] ]
+; SINK-GATHER-NEXT: [[TMP56:%.*]] = phi <8 x i32> [ [[TMP48]], [[PRED_UDIV_CONTINUE10]] ], [ [[TMP54]], [[PRED_UDIV_IF11]] ]
+; SINK-GATHER-NEXT: [[TMP57:%.*]] = extractelement <8 x i1> [[BROADCAST_SPLAT]], i32 7
+; SINK-GATHER-NEXT: br i1 [[TMP57]], label [[PRED_UDIV_IF13:%.*]], label [[PRED_UDIV_CONTINUE14]]
+; SINK-GATHER: pred.udiv.if13:
+; SINK-GATHER-NEXT: [[TMP58:%.*]] = extractelement <8 x i64> [[TMP0]], i32 7
+; SINK-GATHER-NEXT: [[TMP59:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP58]]
+; SINK-GATHER-NEXT: [[TMP60:%.*]] = load i32, i32* [[TMP59]], align 4
+; SINK-GATHER-NEXT: [[TMP61:%.*]] = udiv i32 [[TMP60]], [[X]]
+; SINK-GATHER-NEXT: [[TMP62:%.*]] = insertelement <8 x i32> [[TMP56]], i32 [[TMP61]], i32 7
+; SINK-GATHER-NEXT: br label [[PRED_UDIV_CONTINUE14]]
+; SINK-GATHER: pred.udiv.continue14:
+; SINK-GATHER-NEXT: [[TMP63:%.*]] = phi i32 [ poison, [[PRED_UDIV_CONTINUE12]] ], [ [[TMP60]], [[PRED_UDIV_IF13]] ]
+; SINK-GATHER-NEXT: [[TMP64:%.*]] = phi <8 x i32> [ [[TMP56]], [[PRED_UDIV_CONTINUE12]] ], [ [[TMP62]], [[PRED_UDIV_IF13]] ]
+; SINK-GATHER-NEXT: [[TMP65:%.*]] = xor <8 x i1> [[BROADCAST_SPLAT]], <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>
+; SINK-GATHER-NEXT: [[PREDPHI:%.*]] = select <8 x i1> [[BROADCAST_SPLAT]], <8 x i32> [[TMP64]], <8 x i32> [[BROADCAST_SPLAT16]]
+; SINK-GATHER-NEXT: [[TMP66]] = add <8 x i32> [[VEC_PHI]], [[PREDPHI]]
+; SINK-GATHER-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
+; SINK-GATHER-NEXT: [[VEC_IND_NEXT]] = add <8 x i64> [[VEC_IND]], <i64 8, i64 8, i64 8, i64 8, i64 8, i64 8, i64 8, i64 8>
+; SINK-GATHER-NEXT: [[TMP67:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; SINK-GATHER-NEXT: br i1 [[TMP67]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; SINK-GATHER: middle.block:
+; SINK-GATHER-NEXT: [[TMP68:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP66]])
+; SINK-GATHER-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[SMAX]], [[N_VEC]]
+; SINK-GATHER-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
+; SINK-GATHER: scalar.ph:
+; SINK-GATHER-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; SINK-GATHER-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[TMP68]], [[MIDDLE_BLOCK]] ]
+; SINK-GATHER-NEXT: br label [[FOR_BODY:%.*]]
+; SINK-GATHER: for.body:
+; SINK-GATHER-NEXT: [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ]
+; SINK-GATHER-NEXT: [[R:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[T6:%.*]], [[FOR_INC]] ]
+; SINK-GATHER-NEXT: [[I7:%.*]] = mul i64 [[I]], 777
+; SINK-GATHER-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[FOR_INC]]
+; SINK-GATHER: if.then:
+; SINK-GATHER-NEXT: [[T0:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[I7]]
+; SINK-GATHER-NEXT: [[T2:%.*]] = load i32, i32* [[T0]], align 4
+; SINK-GATHER-NEXT: [[T4:%.*]] = udiv i32 [[T2]], [[X]]
+; SINK-GATHER-NEXT: br label [[FOR_INC]]
+; SINK-GATHER: for.inc:
+; SINK-GATHER-NEXT: [[T5:%.*]] = phi i32 [ [[X]], [[FOR_BODY]] ], [ [[T4]], [[IF_THEN]] ]
+; SINK-GATHER-NEXT: [[T6]] = add i32 [[R]], [[T5]]
+; SINK-GATHER-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
+; SINK-GATHER-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
+; SINK-GATHER-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP5:![0-9]+]]
+; SINK-GATHER: for.end:
+; SINK-GATHER-NEXT: [[T7:%.*]] = phi i32 [ [[T6]], [[FOR_INC]] ], [ [[TMP68]], [[MIDDLE_BLOCK]] ]
+; SINK-GATHER-NEXT: ret i32 [[T7]]
+;
entry:
br label %for.body
for.body:
%i = phi i64 [ 0, %entry ], [ %i.next, %for.inc ]
- %r = phi i32 [ 0, %entry ], [ %tmp6, %for.inc ]
+ %r = phi i32 [ 0, %entry ], [ %t6, %for.inc ]
%i7 = mul i64 %i, 777
br i1 %c, label %if.then, label %for.inc
if.then:
- %tmp0 = getelementptr inbounds i32, i32* %a, i64 %i7
- %tmp2 = load i32, i32* %tmp0, align 4
- %tmp4 = udiv i32 %tmp2, %x
+ %t0 = getelementptr inbounds i32, i32* %a, i64 %i7
+ %t2 = load i32, i32* %t0, align 4
+ %t4 = udiv i32 %t2, %x
br label %for.inc
for.inc:
- %tmp5 = phi i32 [ %x, %for.body ], [ %tmp4, %if.then]
- %tmp6 = add i32 %r, %tmp5
+ %t5 = phi i32 [ %x, %for.body ], [ %t4, %if.then]
+ %t6 = add i32 %r, %t5
%i.next = add nuw nsw i64 %i, 1
%cond = icmp slt i64 %i.next, %n
br i1 %cond, label %for.body, label %for.end
for.end:
- %tmp7 = phi i32 [ %tmp6, %for.inc ]
- ret i32 %tmp7
+ %t7 = phi i32 [ %t6, %for.inc ]
+ ret i32 %t7
}
diff --git a/llvm/test/Transforms/LoopVectorize/induction.ll b/llvm/test/Transforms/LoopVectorize/induction.ll
index 92c360b077934..64e1f09de8e1f 100644
--- a/llvm/test/Transforms/LoopVectorize/induction.ll
+++ b/llvm/test/Transforms/LoopVectorize/induction.ll
@@ -3554,7 +3554,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[N_MOD_VF]]
; CHECK-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; CHECK-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[CAST_VTC]]
-; CHECK-NEXT: [[IND_END3:%.*]] = add i32 [[EXT]], [[N_VEC]]
+; CHECK-NEXT: [[IND_END2:%.*]] = add i32 [[EXT]], [[N_VEC]]
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT]], i32 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[INDUCTION:%.*]] = add <2 x i32> [[DOTSPLAT]], <i32 0, i32 1>
@@ -3579,12 +3579,12 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
; CHECK-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END3]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
+; CHECK-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i8 [[IDX]]
; CHECK-NEXT: store i32 [[SPHI]], i32* [[PTR]], align 4
; CHECK-NEXT: [[IDX_INC]] = add i8 [[IDX]], 1
@@ -3621,7 +3621,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; IND-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], -2
; IND-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; IND-NEXT: [[IND_END:%.*]] = add i8 [[CAST_VTC]], [[T]]
-; IND-NEXT: [[IND_END3:%.*]] = add i32 [[N_VEC]], [[EXT]]
+; IND-NEXT: [[IND_END2:%.*]] = add i32 [[N_VEC]], [[EXT]]
; IND-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT]], i64 0
; IND-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; IND-NEXT: [[INDUCTION:%.*]] = add nuw nsw <2 x i32> [[DOTSPLAT]], <i32 0, i32 1>
@@ -3645,12 +3645,12 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; IND: scalar.ph:
; IND-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
; IND-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
-; IND-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END3]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
+; IND-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
; IND-NEXT: br label [[LOOP:%.*]]
; IND: loop:
; IND-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; IND-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; IND-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
+; IND-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; IND-NEXT: [[TMP15:%.*]] = sext i8 [[IDX]] to i64
; IND-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP15]]
; IND-NEXT: store i32 [[SPHI]], i32* [[PTR]], align 4
@@ -3688,7 +3688,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; UNROLL-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], -4
; UNROLL-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; UNROLL-NEXT: [[IND_END:%.*]] = add i8 [[CAST_VTC]], [[T]]
-; UNROLL-NEXT: [[IND_END3:%.*]] = add i32 [[N_VEC]], [[EXT]]
+; UNROLL-NEXT: [[IND_END2:%.*]] = add i32 [[N_VEC]], [[EXT]]
; UNROLL-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT]], i64 0
; UNROLL-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; UNROLL-NEXT: [[INDUCTION:%.*]] = add nuw nsw <2 x i32> [[DOTSPLAT]], <i32 0, i32 1>
@@ -3716,12 +3716,12 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; UNROLL: scalar.ph:
; UNROLL-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
; UNROLL-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
-; UNROLL-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END3]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
+; UNROLL-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
; UNROLL-NEXT: br label [[LOOP:%.*]]
; UNROLL: loop:
; UNROLL-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; UNROLL-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
+; UNROLL-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[TMP17:%.*]] = sext i8 [[IDX]] to i64
; UNROLL-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP17]]
; UNROLL-NEXT: store i32 [[SPHI]], i32* [[PTR]], align 4
@@ -3763,7 +3763,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; UNROLL-NO-IC-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[N_MOD_VF]]
; UNROLL-NO-IC-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; UNROLL-NO-IC-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[CAST_VTC]]
-; UNROLL-NO-IC-NEXT: [[IND_END3:%.*]] = add i32 [[EXT]], [[N_VEC]]
+; UNROLL-NO-IC-NEXT: [[IND_END2:%.*]] = add i32 [[EXT]], [[N_VEC]]
; UNROLL-NO-IC-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT]], i32 0
; UNROLL-NO-IC-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; UNROLL-NO-IC-NEXT: [[INDUCTION:%.*]] = add <2 x i32> [[DOTSPLAT]], <i32 0, i32 1>
@@ -3794,12 +3794,12 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; UNROLL-NO-IC: scalar.ph:
; UNROLL-NO-IC-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
; UNROLL-NO-IC-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
-; UNROLL-NO-IC-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END3]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
+; UNROLL-NO-IC-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
; UNROLL-NO-IC-NEXT: br label [[LOOP:%.*]]
; UNROLL-NO-IC: loop:
; UNROLL-NO-IC-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NO-IC-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; UNROLL-NO-IC-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
+; UNROLL-NO-IC-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; UNROLL-NO-IC-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i8 [[IDX]]
; UNROLL-NO-IC-NEXT: store i32 [[SPHI]], i32* [[PTR]], align 4
; UNROLL-NO-IC-NEXT: [[IDX_INC]] = add i8 [[IDX]], 1
@@ -3836,7 +3836,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], -8
; INTERLEAVE-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i8 [[CAST_VTC]], [[T]]
-; INTERLEAVE-NEXT: [[IND_END3:%.*]] = add i32 [[N_VEC]], [[EXT]]
+; INTERLEAVE-NEXT: [[IND_END2:%.*]] = add i32 [[N_VEC]], [[EXT]]
; INTERLEAVE-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[EXT]], i64 0
; INTERLEAVE-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; INTERLEAVE-NEXT: [[INDUCTION:%.*]] = add nuw nsw <4 x i32> [[DOTSPLAT]], <i32 0, i32 1, i32 2, i32 3>
@@ -3864,12 +3864,12 @@ define void @wrappingindvars1(i8 %t, i32 %len, i32 *%A) {
; INTERLEAVE: scalar.ph:
; INTERLEAVE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
; INTERLEAVE-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
-; INTERLEAVE-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END3]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
+; INTERLEAVE-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT]], [[LOOP_PREHEADER]] ], [ [[EXT]], [[VECTOR_SCEVCHECK]] ]
; INTERLEAVE-NEXT: br label [[LOOP:%.*]]
; INTERLEAVE: loop:
; INTERLEAVE-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; INTERLEAVE-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
+; INTERLEAVE-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[TMP17:%.*]] = sext i8 [[IDX]] to i64
; INTERLEAVE-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP17]]
; INTERLEAVE-NEXT: store i32 [[SPHI]], i32* [[PTR]], align 4
@@ -3944,7 +3944,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; CHECK-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; CHECK-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[CAST_VTC]]
; CHECK-NEXT: [[TMP12:%.*]] = mul i32 [[N_VEC]], 4
-; CHECK-NEXT: [[IND_END2:%.*]] = add i32 [[EXT_MUL]], [[TMP12]]
+; CHECK-NEXT: [[IND_END1:%.*]] = add i32 [[EXT_MUL]], [[TMP12]]
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT_MUL]], i32 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[INDUCTION:%.*]] = add <2 x i32> [[DOTSPLAT]], <i32 0, i32 4>
@@ -3968,12 +3968,12 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; CHECK-NEXT: br i1 [[CMP_N]], label [[EXIT_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END1]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
; CHECK-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
+; CHECK-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i8 [[IDX]]
; CHECK-NEXT: store i32 [[SPHI]], i32* [[PTR]], align 4
@@ -4014,7 +4014,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; IND-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; IND-NEXT: [[IND_END:%.*]] = add i8 [[CAST_VTC]], [[T]]
; IND-NEXT: [[TMP10:%.*]] = add i32 [[N_VEC]], [[EXT]]
-; IND-NEXT: [[IND_END2:%.*]] = shl i32 [[TMP10]], 2
+; IND-NEXT: [[IND_END1:%.*]] = shl i32 [[TMP10]], 2
; IND-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT_MUL]], i64 0
; IND-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; IND-NEXT: [[INDUCTION:%.*]] = add nuw nsw <2 x i32> [[DOTSPLAT]], <i32 0, i32 4>
@@ -4037,12 +4037,12 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; IND-NEXT: br i1 [[CMP_N]], label [[EXIT_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; IND: scalar.ph:
; IND-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
-; IND-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
+; IND-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END1]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
; IND-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
; IND-NEXT: br label [[LOOP:%.*]]
; IND: loop:
; IND-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; IND-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
+; IND-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; IND-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; IND-NEXT: [[TMP16:%.*]] = sext i8 [[IDX]] to i64
; IND-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP16]]
@@ -4084,7 +4084,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; UNROLL-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; UNROLL-NEXT: [[IND_END:%.*]] = add i8 [[CAST_VTC]], [[T]]
; UNROLL-NEXT: [[TMP10:%.*]] = add i32 [[N_VEC]], [[EXT]]
-; UNROLL-NEXT: [[IND_END2:%.*]] = shl i32 [[TMP10]], 2
+; UNROLL-NEXT: [[IND_END1:%.*]] = shl i32 [[TMP10]], 2
; UNROLL-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT_MUL]], i64 0
; UNROLL-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; UNROLL-NEXT: [[INDUCTION:%.*]] = add nuw nsw <2 x i32> [[DOTSPLAT]], <i32 0, i32 4>
@@ -4111,12 +4111,12 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; UNROLL-NEXT: br i1 [[CMP_N]], label [[EXIT_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; UNROLL: scalar.ph:
; UNROLL-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
-; UNROLL-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
+; UNROLL-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END1]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
; UNROLL-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
; UNROLL-NEXT: br label [[LOOP:%.*]]
; UNROLL: loop:
; UNROLL-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; UNROLL-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
+; UNROLL-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[TMP18:%.*]] = sext i8 [[IDX]] to i64
; UNROLL-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP18]]
@@ -4162,7 +4162,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; UNROLL-NO-IC-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; UNROLL-NO-IC-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[CAST_VTC]]
; UNROLL-NO-IC-NEXT: [[TMP12:%.*]] = mul i32 [[N_VEC]], 4
-; UNROLL-NO-IC-NEXT: [[IND_END2:%.*]] = add i32 [[EXT_MUL]], [[TMP12]]
+; UNROLL-NO-IC-NEXT: [[IND_END1:%.*]] = add i32 [[EXT_MUL]], [[TMP12]]
; UNROLL-NO-IC-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT_MUL]], i32 0
; UNROLL-NO-IC-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; UNROLL-NO-IC-NEXT: [[INDUCTION:%.*]] = add <2 x i32> [[DOTSPLAT]], <i32 0, i32 4>
@@ -4192,12 +4192,12 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; UNROLL-NO-IC-NEXT: br i1 [[CMP_N]], label [[EXIT_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; UNROLL-NO-IC: scalar.ph:
; UNROLL-NO-IC-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
-; UNROLL-NO-IC-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
+; UNROLL-NO-IC-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END1]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
; UNROLL-NO-IC-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
; UNROLL-NO-IC-NEXT: br label [[LOOP:%.*]]
; UNROLL-NO-IC: loop:
; UNROLL-NO-IC-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; UNROLL-NO-IC-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
+; UNROLL-NO-IC-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; UNROLL-NO-IC-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; UNROLL-NO-IC-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i8 [[IDX]]
; UNROLL-NO-IC-NEXT: store i32 [[SPHI]], i32* [[PTR]], align 4
@@ -4238,7 +4238,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; INTERLEAVE-NEXT: [[CAST_VTC:%.*]] = trunc i32 [[N_VEC]] to i8
; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i8 [[CAST_VTC]], [[T]]
; INTERLEAVE-NEXT: [[TMP10:%.*]] = add i32 [[N_VEC]], [[EXT]]
-; INTERLEAVE-NEXT: [[IND_END2:%.*]] = shl i32 [[TMP10]], 2
+; INTERLEAVE-NEXT: [[IND_END1:%.*]] = shl i32 [[TMP10]], 2
; INTERLEAVE-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[EXT_MUL]], i64 0
; INTERLEAVE-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; INTERLEAVE-NEXT: [[INDUCTION:%.*]] = add nuw nsw <4 x i32> [[DOTSPLAT]], <i32 0, i32 4, i32 8, i32 12>
@@ -4265,12 +4265,12 @@ define void @wrappingindvars2(i8 %t, i32 %len, i32 *%A) {
; INTERLEAVE-NEXT: br i1 [[CMP_N]], label [[EXIT_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; INTERLEAVE: scalar.ph:
; INTERLEAVE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i8 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[T]], [[LOOP_PREHEADER]] ], [ [[T]], [[VECTOR_SCEVCHECK]] ]
-; INTERLEAVE-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i32 [ [[IND_END2]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
+; INTERLEAVE-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i32 [ [[IND_END1]], [[MIDDLE_BLOCK]] ], [ [[EXT_MUL]], [[LOOP_PREHEADER]] ], [ [[EXT_MUL]], [[VECTOR_SCEVCHECK]] ]
; INTERLEAVE-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_PREHEADER]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
; INTERLEAVE-NEXT: br label [[LOOP:%.*]]
; INTERLEAVE: loop:
; INTERLEAVE-NEXT: [[IDX:%.*]] = phi i8 [ [[IDX_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; INTERLEAVE-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
+; INTERLEAVE-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[TMP18:%.*]] = sext i8 [[IDX]] to i64
; INTERLEAVE-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP18]]
diff --git a/llvm/test/Transforms/LoopVectorize/pr44488-predication.ll b/llvm/test/Transforms/LoopVectorize/pr44488-predication.ll
index 219409685dd6e..f78ef3ee4c426 100644
--- a/llvm/test/Transforms/LoopVectorize/pr44488-predication.ll
+++ b/llvm/test/Transforms/LoopVectorize/pr44488-predication.ll
@@ -15,36 +15,36 @@ define i16 @test_true_and_false_branch_equal() {
; CHECK: vector.ph:
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_SREM_CONTINUE4:%.*]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = load i16, i16* @v_38, align 1
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <2 x i16> poison, i16 [[TMP2]], i32 0
-; CHECK-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <2 x i16> [[BROADCAST_SPLATINSERT1]], <2 x i16> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <2 x i16> [[BROADCAST_SPLAT2]], zeroinitializer
-; CHECK-NEXT: [[TMP5:%.*]] = xor <2 x i1> [[TMP4]], <i1 true, i1 true>
-; CHECK-NEXT: [[TMP6:%.*]] = extractelement <2 x i1> [[TMP5]], i32 0
-; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_SREM_IF:%.*]], label [[PRED_SREM_CONTINUE:%.*]]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_SREM_CONTINUE2:%.*]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = load i16, i16* @v_38, align 1
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i16> poison, i16 [[TMP0]], i32 0
+; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i16> [[BROADCAST_SPLATINSERT]], <2 x i16> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = icmp eq <2 x i16> [[BROADCAST_SPLAT]], zeroinitializer
+; CHECK-NEXT: [[TMP2:%.*]] = xor <2 x i1> [[TMP1]], <i1 true, i1 true>
+; CHECK-NEXT: [[TMP3:%.*]] = extractelement <2 x i1> [[TMP2]], i32 0
+; CHECK-NEXT: br i1 [[TMP3]], label [[PRED_SREM_IF:%.*]], label [[PRED_SREM_CONTINUE:%.*]]
; CHECK: pred.srem.if:
-; CHECK-NEXT: [[TMP7:%.*]] = srem i16 5786, [[TMP2]]
-; CHECK-NEXT: [[TMP8:%.*]] = insertelement <2 x i16> poison, i16 [[TMP7]], i32 0
+; CHECK-NEXT: [[TMP4:%.*]] = srem i16 5786, [[TMP0]]
+; CHECK-NEXT: [[TMP5:%.*]] = insertelement <2 x i16> poison, i16 [[TMP4]], i32 0
; CHECK-NEXT: br label [[PRED_SREM_CONTINUE]]
; CHECK: pred.srem.continue:
-; CHECK-NEXT: [[TMP9:%.*]] = phi <2 x i16> [ poison, [[VECTOR_BODY]] ], [ [[TMP8]], [[PRED_SREM_IF]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = extractelement <2 x i1> [[TMP5]], i32 1
-; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_SREM_IF3:%.*]], label [[PRED_SREM_CONTINUE4]]
+; CHECK-NEXT: [[TMP6:%.*]] = phi <2 x i16> [ poison, [[VECTOR_BODY]] ], [ [[TMP5]], [[PRED_SREM_IF]] ]
+; CHECK-NEXT: [[TMP7:%.*]] = extractelement <2 x i1> [[TMP2]], i32 1
+; CHECK-NEXT: br i1 [[TMP7]], label [[PRED_SREM_IF1:%.*]], label [[PRED_SREM_CONTINUE2]]
; CHECK: pred.srem.if1:
-; CHECK-NEXT: [[TMP11:%.*]] = srem i16 5786, [[TMP2]]
-; CHECK-NEXT: [[TMP12:%.*]] = insertelement <2 x i16> [[TMP9]], i16 [[TMP11]], i32 1
-; CHECK-NEXT: br label [[PRED_SREM_CONTINUE4]]
+; CHECK-NEXT: [[TMP8:%.*]] = srem i16 5786, [[TMP0]]
+; CHECK-NEXT: [[TMP9:%.*]] = insertelement <2 x i16> [[TMP6]], i16 [[TMP8]], i32 1
+; CHECK-NEXT: br label [[PRED_SREM_CONTINUE2]]
; CHECK: pred.srem.continue2:
-; CHECK-NEXT: [[TMP13:%.*]] = phi <2 x i16> [ [[TMP9]], [[PRED_SREM_CONTINUE]] ], [ [[TMP12]], [[PRED_SREM_IF3]] ]
-; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP4]], <2 x i16> <i16 5786, i16 5786>, <2 x i16> [[TMP13]]
-; CHECK-NEXT: [[TMP14:%.*]] = extractelement <2 x i16> [[PREDPHI]], i32 0
-; CHECK-NEXT: store i16 [[TMP14]], i16* @v_39, align 1
-; CHECK-NEXT: [[TMP15:%.*]] = extractelement <2 x i16> [[PREDPHI]], i32 1
-; CHECK-NEXT: store i16 [[TMP15]], i16* @v_39, align 1
+; CHECK-NEXT: [[TMP10:%.*]] = phi <2 x i16> [ [[TMP6]], [[PRED_SREM_CONTINUE]] ], [ [[TMP9]], [[PRED_SREM_IF1]] ]
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP1]], <2 x i16> <i16 5786, i16 5786>, <2 x i16> [[TMP10]]
+; CHECK-NEXT: [[TMP11:%.*]] = extractelement <2 x i16> [[PREDPHI]], i32 0
+; CHECK-NEXT: store i16 [[TMP11]], i16* @v_39, align 1
+; CHECK-NEXT: [[TMP12:%.*]] = extractelement <2 x i16> [[PREDPHI]], i32 1
+; CHECK-NEXT: store i16 [[TMP12]], i16* @v_39, align 1
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
-; CHECK-NEXT: [[TMP16:%.*]] = icmp eq i32 [[INDEX_NEXT]], 12
-; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], [[LOOP0:!llvm.loop !.*]]
+; CHECK-NEXT: [[TMP13:%.*]] = icmp eq i32 [[INDEX_NEXT]], 12
+; CHECK-NEXT: br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 12, 12
; CHECK-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
@@ -67,7 +67,7 @@ define i16 @test_true_and_false_branch_equal() {
; CHECK-NEXT: store i16 [[COND6]], i16* @v_39, align 1
; CHECK-NEXT: [[INC7]] = add nsw i16 [[I_07]], 1
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i16 [[INC7]], 111
-; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[EXIT]], [[LOOP2:!llvm.loop !.*]]
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[EXIT]], !llvm.loop [[LOOP2:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: [[RV:%.*]] = load i16, i16* @v_39, align 1
; CHECK-NEXT: ret i16 [[RV]]
diff --git a/llvm/test/Transforms/LoopVectorize/reduction-small-size.ll b/llvm/test/Transforms/LoopVectorize/reduction-small-size.ll
index 9075478b4b235..eba154e0b5b99 100644
--- a/llvm/test/Transforms/LoopVectorize/reduction-small-size.ll
+++ b/llvm/test/Transforms/LoopVectorize/reduction-small-size.ll
@@ -1,20 +1,87 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -force-vector-width=4 -force-vector-interleave=1 -passes=loop-vectorize -S | FileCheck %s
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+define i8 @PR34687(i1 %c, i32 %x, i32 %n) {
; CHECK-LABEL: @PR34687(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N:%.*]], 4
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK: vector.ph:
+; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[N]], 4
+; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[N]], [[N_MOD_VF]]
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[C:%.*]], i32 0
+; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT7:%.*]] = insertelement <4 x i32> poison, i32 [[X:%.*]], i32 0
+; CHECK-NEXT: [[BROADCAST_SPLAT8:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT7]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, %vector.ph ], [ [[INDEX_NEXT:%.*]], %[[LATCH:.*]] ]
-; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, %vector.ph ], [ [[TMP17:%.*]], %[[LATCH]] ]
-; CHECK: [[LATCH]]:
-; CHECK: [[TMP13:%.*]] = and <4 x i32> [[VEC_PHI]], <i32 255, i32 255, i32 255, i32 255>
-; CHECK-NEXT: [[TMP14:%.*]] = add <4 x i32> [[TMP13]], {{.*}}
+; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_SDIV_CONTINUE6:%.*]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP16:%.*]], [[PRED_SDIV_CONTINUE6]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = extractelement <4 x i1> [[BROADCAST_SPLAT]], i32 0
+; CHECK-NEXT: br i1 [[TMP0]], label [[PRED_SDIV_IF:%.*]], label [[PRED_SDIV_CONTINUE:%.*]]
+; CHECK: pred.sdiv.if:
+; CHECK-NEXT: [[TMP1:%.*]] = sdiv i32 undef, undef
+; CHECK-NEXT: br label [[PRED_SDIV_CONTINUE]]
+; CHECK: pred.sdiv.continue:
+; CHECK-NEXT: [[TMP2:%.*]] = phi i32 [ poison, [[VECTOR_BODY]] ], [ [[TMP1]], [[PRED_SDIV_IF]] ]
+; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x i1> [[BROADCAST_SPLAT]], i32 1
+; CHECK-NEXT: br i1 [[TMP3]], label [[PRED_SDIV_IF1:%.*]], label [[PRED_SDIV_CONTINUE2:%.*]]
+; CHECK: pred.sdiv.if1:
+; CHECK-NEXT: [[TMP4:%.*]] = sdiv i32 undef, undef
+; CHECK-NEXT: br label [[PRED_SDIV_CONTINUE2]]
+; CHECK: pred.sdiv.continue2:
+; CHECK-NEXT: [[TMP5:%.*]] = phi i32 [ poison, [[PRED_SDIV_CONTINUE]] ], [ [[TMP4]], [[PRED_SDIV_IF1]] ]
+; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[BROADCAST_SPLAT]], i32 2
+; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_SDIV_IF3:%.*]], label [[PRED_SDIV_CONTINUE4:%.*]]
+; CHECK: pred.sdiv.if3:
+; CHECK-NEXT: [[TMP7:%.*]] = sdiv i32 undef, undef
+; CHECK-NEXT: br label [[PRED_SDIV_CONTINUE4]]
+; CHECK: pred.sdiv.continue4:
+; CHECK-NEXT: [[TMP8:%.*]] = phi i32 [ poison, [[PRED_SDIV_CONTINUE2]] ], [ [[TMP7]], [[PRED_SDIV_IF3]] ]
+; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[BROADCAST_SPLAT]], i32 3
+; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_SDIV_IF5:%.*]], label [[PRED_SDIV_CONTINUE6]]
+; CHECK: pred.sdiv.if5:
+; CHECK-NEXT: [[TMP10:%.*]] = sdiv i32 undef, undef
+; CHECK-NEXT: br label [[PRED_SDIV_CONTINUE6]]
+; CHECK: pred.sdiv.continue6:
+; CHECK-NEXT: [[TMP11:%.*]] = phi i32 [ poison, [[PRED_SDIV_CONTINUE4]] ], [ [[TMP10]], [[PRED_SDIV_IF5]] ]
+; CHECK-NEXT: [[TMP12:%.*]] = and <4 x i32> [[VEC_PHI]], <i32 255, i32 255, i32 255, i32 255>
+; CHECK-NEXT: [[TMP13:%.*]] = add <4 x i32> [[TMP12]], [[BROADCAST_SPLAT8]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
-; CHECK: [[TMP16:%.*]] = trunc <4 x i32> [[TMP14]] to <4 x i8>
-; CHECK-NEXT: [[TMP17]] = zext <4 x i8> [[TMP16]] to <4 x i32>
-; CHECK-NEXT: br i1 {{.*}}, label %middle.block, label %vector.body
+; CHECK-NEXT: [[TMP14:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: [[TMP15:%.*]] = trunc <4 x i32> [[TMP13]] to <4 x i8>
+; CHECK-NEXT: [[TMP16]] = zext <4 x i8> [[TMP15]] to <4 x i32>
+; CHECK-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: middle.block:
+; CHECK-NEXT: [[TMP17:%.*]] = trunc <4 x i32> [[TMP16]] to <4 x i8>
+; CHECK-NEXT: [[TMP18:%.*]] = call i8 @llvm.vector.reduce.add.v4i8(<4 x i8> [[TMP17]])
+; CHECK-NEXT: [[TMP19:%.*]] = zext i8 [[TMP18]] to i32
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
+; CHECK: scalar.ph:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[TMP19]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[IF_END:%.*]] ]
+; CHECK-NEXT: [[R:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[R_NEXT:%.*]], [[IF_END]] ]
+; CHECK-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[IF_END]]
+; CHECK: if.then:
+; CHECK-NEXT: [[T0:%.*]] = sdiv i32 undef, undef
+; CHECK-NEXT: br label [[IF_END]]
+; CHECK: if.end:
+; CHECK-NEXT: [[T1:%.*]] = and i32 [[R]], 255
+; CHECK-NEXT: [[I_NEXT]] = add nsw i32 [[I]], 1
+; CHECK-NEXT: [[R_NEXT]] = add nuw nsw i32 [[T1]], [[X]]
+; CHECK-NEXT: [[COND:%.*]] = icmp eq i32 [[I_NEXT]], [[N]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
+; CHECK: for.end:
+; CHECK-NEXT: [[T2:%.*]] = phi i32 [ [[R_NEXT]], [[IF_END]] ], [ [[TMP19]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: [[T3:%.*]] = trunc i32 [[T2]] to i8
+; CHECK-NEXT: ret i8 [[T3]]
;
-define i8 @PR34687(i1 %c, i32 %x, i32 %n) {
entry:
br label %for.body
@@ -24,57 +91,100 @@ for.body:
br i1 %c, label %if.then, label %if.end
if.then:
- %tmp0 = sdiv i32 undef, undef
+ %t0 = sdiv i32 undef, undef
br label %if.end
if.end:
- %tmp1 = and i32 %r, 255
+ %t1 = and i32 %r, 255
%i.next = add nsw i32 %i, 1
- %r.next = add nuw nsw i32 %tmp1, %x
+ %r.next = add nuw nsw i32 %t1, %x
%cond = icmp eq i32 %i.next, %n
br i1 %cond, label %for.end, label %for.body
for.end:
- %tmp2 = phi i32 [ %r.next, %if.end ]
- %tmp3 = trunc i32 %tmp2 to i8
- ret i8 %tmp3
+ %t2 = phi i32 [ %r.next, %if.end ]
+ %t3 = trunc i32 %t2 to i8
+ ret i8 %t3
}
+define i32 @PR35734(i32 %x, i32 %y) {
; CHECK-LABEL: @PR35734(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[X:%.*]], i32 78)
+; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[SMAX]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = sub i32 [[TMP0]], [[X]]
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP1]], 4
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK: [[TMP3:%.*]] = insertelement <4 x i32> zeroinitializer, i32 %y, i32 0
-; CHECK-NEXT: br label %vector.body
+; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[TMP1]], 4
+; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP1]], [[N_MOD_VF]]
+; CHECK-NEXT: [[IND_END:%.*]] = add i32 [[X]], [[N_VEC]]
+; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[Y:%.*]], i32 0
+; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, %vector.ph ], [ [[INDEX_NEXT:%.*]], %vector.body ]
-; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP3]], %vector.ph ], [ [[TMP9:%.*]], %vector.body ]
-; CHECK: [[TMP5:%.*]] = and <4 x i32> [[VEC_PHI]], <i32 1, i32 1, i32 1, i32 1>
-; CHECK-NEXT: [[TMP6:%.*]] = add <4 x i32> [[TMP5]], <i32 -1, i32 -1, i32 -1, i32 -1>
+; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP2]], [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP3:%.*]] = and <4 x i32> [[VEC_PHI]], <i32 1, i32 1, i32 1, i32 1>
+; CHECK-NEXT: [[TMP4:%.*]] = add <4 x i32> [[TMP3]], <i32 -1, i32 -1, i32 -1, i32 -1>
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
-; CHECK: [[TMP8:%.*]] = trunc <4 x i32> [[TMP6]] to <4 x i1>
-; CHECK-NEXT: [[TMP9]] = sext <4 x i1> [[TMP8]] to <4 x i32>
-; CHECK-NEXT: br i1 {{.*}}, label %middle.block, label %vector.body
+; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: [[TMP6:%.*]] = trunc <4 x i32> [[TMP4]] to <4 x i1>
+; CHECK-NEXT: [[TMP7]] = sext <4 x i1> [[TMP6]] to <4 x i32>
+; CHECK-NEXT: br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK: middle.block:
+; CHECK-NEXT: [[TMP8:%.*]] = trunc <4 x i32> [[TMP7]] to <4 x i1>
+; CHECK-NEXT: [[TMP9:%.*]] = call i1 @llvm.vector.reduce.add.v4i1(<4 x i1> [[TMP8]])
+; CHECK-NEXT: [[TMP10:%.*]] = sext i1 [[TMP9]] to i32
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[TMP1]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
+; CHECK: scalar.ph:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[X]], [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ [[Y]], [[ENTRY]] ], [ [[TMP10]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[FOR_BODY]] ]
+; CHECK-NEXT: [[R:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[R_NEXT:%.*]], [[FOR_BODY]] ]
+; CHECK-NEXT: [[T0:%.*]] = and i32 [[R]], 1
+; CHECK-NEXT: [[R_NEXT]] = add i32 [[T0]], -1
+; CHECK-NEXT: [[I_NEXT]] = add nsw i32 [[I]], 1
+; CHECK-NEXT: [[COND:%.*]] = icmp sgt i32 [[I]], 77
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK: for.end:
+; CHECK-NEXT: [[T1:%.*]] = phi i32 [ [[R_NEXT]], [[FOR_BODY]] ], [ [[TMP10]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: ret i32 [[T1]]
;
-define i32 @PR35734(i32 %x, i32 %y) {
entry:
br label %for.body
for.body:
%i = phi i32 [ %x, %entry ], [ %i.next, %for.body ]
%r = phi i32 [ %y, %entry ], [ %r.next, %for.body ]
- %tmp0 = and i32 %r, 1
- %r.next = add i32 %tmp0, -1
+ %t0 = and i32 %r, 1
+ %r.next = add i32 %t0, -1
%i.next = add nsw i32 %i, 1
%cond = icmp sgt i32 %i, 77
br i1 %cond, label %for.end, label %for.body
for.end:
- %tmp1 = phi i32 [ %r.next, %for.body ]
- ret i32 %tmp1
+ %t1 = phi i32 [ %r.next, %for.body ]
+ ret i32 %t1
}
define i32 @pr51794_signed_negative(i16 %iv.start, i32 %xor.start) {
-; CHECK-LABEL: define {{.*}} @pr51794_signed_negative(
-; CHECK-NOT: vector.body:
+; CHECK-LABEL: @pr51794_signed_negative(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[LOOP:%.*]]
+; CHECK: loop:
+; CHECK-NEXT: [[XOR_RED:%.*]] = phi i32 [ [[XOR_START:%.*]], [[ENTRY:%.*]] ], [ [[XOR:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[IV:%.*]] = phi i16 [ [[IV_START:%.*]], [[ENTRY]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[IV_NEXT]] = add i16 [[IV]], -1
+; CHECK-NEXT: [[AND:%.*]] = and i32 [[XOR_RED]], 1
+; CHECK-NEXT: [[XOR]] = xor i32 [[AND]], -1
+; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i16 [[IV_NEXT]], 0
+; CHECK-NEXT: br i1 [[TOBOOL_NOT]], label [[EXIT:%.*]], label [[LOOP]]
+; CHECK: exit:
+; CHECK-NEXT: [[XOR_LCSSA:%.*]] = phi i32 [ [[XOR]], [[LOOP]] ]
+; CHECK-NEXT: ret i32 [[XOR_LCSSA]]
;
entry:
br label %loop
@@ -94,8 +204,20 @@ exit:
}
define i32 @pr52485_signed_negative(i32 %xor.start) {
-; CHECK-LABEL: define {{.*}} @pr52485_signed_negative(
-; CHECK-NOT: vector.body:
+; CHECK-LABEL: @pr52485_signed_negative(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[LOOP:%.*]]
+; CHECK: loop:
+; CHECK-NEXT: [[IV:%.*]] = phi i32 [ -23, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[XOR_RED:%.*]] = phi i32 [ [[XOR_START:%.*]], [[ENTRY]] ], [ [[XOR:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[AND:%.*]] = and i32 [[XOR_RED]], 255
+; CHECK-NEXT: [[XOR]] = xor i32 [[AND]], -9
+; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i32 [[IV]], 2
+; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[IV_NEXT]], -15
+; CHECK-NEXT: br i1 [[CMP_NOT]], label [[EXIT:%.*]], label [[LOOP]]
+; CHECK: exit:
+; CHECK-NEXT: [[XOR_LCSSA:%.*]] = phi i32 [ [[XOR]], [[LOOP]] ]
+; CHECK-NEXT: ret i32 [[XOR_LCSSA]]
;
entry:
br label %loop
More information about the llvm-commits
mailing list