Skip to content

[LoopVectorize] Ensure fairness when selecting epilogue VFs #116607

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from
Closed

[LoopVectorize] Ensure fairness when selecting epilogue VFs #116607

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4731,7 +4731,7 @@ VectorizationFactor LoopVectorizationPlanner::selectEpilogueVectorizationFactor(

unsigned Multiplier = IC;
if (MainLoopVF.isScalable())
Multiplier = getVScaleForTuning(OrigLoop, TTI).value_or(1);
Multiplier *= getVScaleForTuning(OrigLoop, TTI).value_or(1);

if (!CM.isEpilogueVectorizationProfitable(MainLoopVF, Multiplier)) {
LLVM_DEBUG(dbgs() << "LEV: Epilogue vectorization is not profitable for "
Expand Down
63 changes: 52 additions & 11 deletions llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -302,10 +302,15 @@ exit:
define i16 @reduce_udiv(ptr %src, i16 %x, i64 %N) #0 {
; DEFAULT-LABEL: define i16 @reduce_udiv(
; DEFAULT-SAME: ptr [[SRC:%.*]], i16 [[X:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; DEFAULT-NEXT: entry:
; DEFAULT-NEXT: iter.check:
; DEFAULT-NEXT: [[TMP0:%.*]] = add i64 [[N]], 1
; DEFAULT-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; DEFAULT-NEXT: [[TMP2:%.*]] = mul i64 [[TMP1]], 8
; DEFAULT-NEXT: [[TMP8:%.*]] = mul i64 [[TMP1]], 2
; DEFAULT-NEXT: [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[TMP0]], [[TMP8]]
; DEFAULT-NEXT: br i1 [[MIN_ITERS_CHECK1]], label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[ENTRY:%.*]]
; DEFAULT: vector.main.loop.iter.check:
; DEFAULT-NEXT: [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
; DEFAULT-NEXT: [[TMP2:%.*]] = mul i64 [[TMP9]], 8
; DEFAULT-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], [[TMP2]]
; DEFAULT-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; DEFAULT: vector.ph:
Expand Down Expand Up @@ -336,28 +341,63 @@ define i16 @reduce_udiv(ptr %src, i16 %x, i64 %N) #0 {
; DEFAULT-NEXT: [[TMP22]] = or <vscale x 4 x i16> [[TMP20]], [[VEC_PHI1]]
; DEFAULT-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; DEFAULT-NEXT: [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; DEFAULT-NEXT: br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; DEFAULT-NEXT: br i1 [[TMP23]], label [[MIDDLE_BLOCK1:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; DEFAULT: middle.block:
; DEFAULT-NEXT: [[BIN_RDX:%.*]] = or <vscale x 4 x i16> [[TMP22]], [[TMP21]]
; DEFAULT-NEXT: [[TMP24:%.*]] = call i16 @llvm.vector.reduce.or.nxv4i16(<vscale x 4 x i16> [[BIN_RDX]])
; 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:%.*]] ]
; DEFAULT-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; DEFAULT: vec.epilog.iter.check:
; DEFAULT-NEXT: [[N_VEC_REMAINING:%.*]] = sub i64 [[TMP0]], [[N_VEC]]
; DEFAULT-NEXT: [[TMP35:%.*]] = call i64 @llvm.vscale.i64()
; DEFAULT-NEXT: [[TMP36:%.*]] = mul i64 [[TMP35]], 2
; DEFAULT-NEXT: [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], [[TMP36]]
; DEFAULT-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH]], label [[SCALAR_PH]]
; DEFAULT: vec.epilog.ph:
; DEFAULT-NEXT: [[BC_MERGE_RDX:%.*]] = phi i16 [ [[TMP24]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
; DEFAULT-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
; DEFAULT-NEXT: [[TMP37:%.*]] = call i64 @llvm.vscale.i64()
; DEFAULT-NEXT: [[TMP38:%.*]] = mul i64 [[TMP37]], 2
; DEFAULT-NEXT: [[N_MOD_VF4:%.*]] = urem i64 [[TMP0]], [[TMP38]]
; DEFAULT-NEXT: [[N_VEC5:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF4]]
; DEFAULT-NEXT: [[TMP25:%.*]] = call i64 @llvm.vscale.i64()
; DEFAULT-NEXT: [[TMP26:%.*]] = mul i64 [[TMP25]], 2
; DEFAULT-NEXT: [[TMP27:%.*]] = insertelement <vscale x 2 x i16> zeroinitializer, i16 [[BC_MERGE_RDX]], i32 0
; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT9:%.*]] = insertelement <vscale x 2 x i16> poison, i16 [[X]], i64 0
; DEFAULT-NEXT: [[BROADCAST_SPLAT10:%.*]] = shufflevector <vscale x 2 x i16> [[BROADCAST_SPLATINSERT9]], <vscale x 2 x i16> poison, <vscale x 2 x i32> zeroinitializer
; DEFAULT-NEXT: br label [[LOOP:%.*]]
; DEFAULT: vec.epilog.vector.body:
; DEFAULT-NEXT: [[INDEX6:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDEX_NEXT11:%.*]], [[LOOP]] ]
; DEFAULT-NEXT: [[VEC_PHI7:%.*]] = phi <vscale x 2 x i16> [ [[TMP27]], [[SCALAR_PH]] ], [ [[TMP32:%.*]], [[LOOP]] ]
; DEFAULT-NEXT: [[TMP28:%.*]] = add i64 [[INDEX6]], 0
; DEFAULT-NEXT: [[TMP29:%.*]] = getelementptr i16, ptr [[SRC]], i64 [[TMP28]]
; DEFAULT-NEXT: [[TMP30:%.*]] = getelementptr i16, ptr [[TMP29]], i32 0
; DEFAULT-NEXT: [[WIDE_LOAD8:%.*]] = load <vscale x 2 x i16>, ptr [[TMP30]], align 2
; DEFAULT-NEXT: [[TMP31:%.*]] = udiv <vscale x 2 x i16> [[WIDE_LOAD8]], [[BROADCAST_SPLAT10]]
; DEFAULT-NEXT: [[TMP32]] = or <vscale x 2 x i16> [[TMP31]], [[VEC_PHI7]]
; DEFAULT-NEXT: [[INDEX_NEXT11]] = add nuw i64 [[INDEX6]], [[TMP26]]
; DEFAULT-NEXT: [[TMP33:%.*]] = icmp eq i64 [[INDEX_NEXT11]], [[N_VEC5]]
; DEFAULT-NEXT: br i1 [[TMP33]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
; DEFAULT: vec.epilog.middle.block:
; DEFAULT-NEXT: [[TMP34:%.*]] = call i16 @llvm.vector.reduce.or.nxv2i16(<vscale x 2 x i16> [[TMP32]])
; DEFAULT-NEXT: [[CMP_N12:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC5]]
; 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_VEC5]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ITER_CHECK:%.*]] ]
; DEFAULT-NEXT: [[BC_MERGE_RDX13:%.*]] = phi i16 [ [[TMP34]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ 0, [[ITER_CHECK]] ], [ [[TMP24]], [[MIDDLE_BLOCK]] ]
; DEFAULT-NEXT: br label [[LOOP1:%.*]]
; DEFAULT: loop:
; DEFAULT-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
; DEFAULT-NEXT: [[RED:%.*]] = phi i16 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[RED_NEXT:%.*]], [[LOOP]] ]
; DEFAULT-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP1]] ]
; DEFAULT-NEXT: [[RED:%.*]] = phi i16 [ [[BC_MERGE_RDX13]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[RED_NEXT:%.*]], [[LOOP1]] ]
; DEFAULT-NEXT: [[GEP:%.*]] = getelementptr i16, ptr [[SRC]], i64 [[IV]]
; DEFAULT-NEXT: [[L:%.*]] = load i16, ptr [[GEP]], align 2
; DEFAULT-NEXT: [[DIV:%.*]] = udiv i16 [[L]], [[X]]
; DEFAULT-NEXT: [[RED_NEXT]] = or i16 [[DIV]], [[RED]]
; DEFAULT-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
; DEFAULT-NEXT: [[EC:%.*]] = icmp eq i64 [[IV]], [[N]]
; DEFAULT-NEXT: br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
; DEFAULT-NEXT: br i1 [[EC]], label [[EXIT]], label [[LOOP1]], !llvm.loop [[LOOP6:![0-9]+]]
; DEFAULT: exit:
; DEFAULT-NEXT: [[RED_NEXT_LCSSA:%.*]] = phi i16 [ [[RED_NEXT]], [[LOOP]] ], [ [[TMP24]], [[MIDDLE_BLOCK]] ]
; DEFAULT-NEXT: [[RED_NEXT_LCSSA:%.*]] = phi i16 [ [[RED_NEXT]], [[LOOP1]] ], [ [[TMP24]], [[MIDDLE_BLOCK1]] ], [ [[TMP34]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
; DEFAULT-NEXT: ret i16 [[RED_NEXT_LCSSA]]
;
; PRED-LABEL: define i16 @reduce_udiv(
Expand Down Expand Up @@ -445,7 +485,8 @@ attributes #0 = { "target-features"="+sve" }
; DEFAULT: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
; DEFAULT: [[LOOP3]] = distinct !{[[LOOP3]], [[META2]], [[META1]]}
; DEFAULT: [[LOOP4]] = distinct !{[[LOOP4]], [[META1]], [[META2]]}
; DEFAULT: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META1]]}
; DEFAULT: [[LOOP5]] = distinct !{[[LOOP5]], [[META1]], [[META2]]}
; DEFAULT: [[LOOP6]] = distinct !{[[LOOP6]], [[META2]], [[META1]]}
;.
; PRED: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
; PRED: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
Expand Down
60 changes: 51 additions & 9 deletions llvm/test/Transforms/LoopVectorize/AArch64/sve-gather-scatter.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -281,10 +281,15 @@ for.cond.cleanup: ; preds = %for.inc, %entry

define void @gather_nxv4i32_ind64_stride2(ptr noalias nocapture %a, ptr noalias nocapture readonly %b, i64 %n) #0 {
; CHECK-LABEL: @gather_nxv4i32_ind64_stride2(
; CHECK-NEXT: entry:
; CHECK-NEXT: iter.check:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 3
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ugt i64 [[N:%.*]], [[TMP1]]
; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP0]], 1
; CHECK-NEXT: [[MIN_ITERS_CHECK_NOT:%.*]] = icmp ugt i64 [[N:%.*]], [[TMP8]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK_NOT]], label [[ENTRY:%.*]], label [[VEC_EPILOG_SCALAR_PH:%.*]]
; CHECK: vector.main.loop.iter.check:
; CHECK-NEXT: [[TMP17:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP17]], 3
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ugt i64 [[N]], [[TMP1]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[VECTOR_PH:%.*]], label [[SCALAR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
Expand Down Expand Up @@ -320,22 +325,59 @@ define void @gather_nxv4i32_ind64_stride2(ptr noalias nocapture %a, ptr noalias
; CHECK-NEXT: store <vscale x 4 x float> [[WIDE_MASKED_GATHER2]], ptr [[TMP14]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP4]]
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC1]]
; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK1:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC1]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[MIDDLE_BLOCK:%.*]]
; CHECK: vec.epilog.iter.check:
; CHECK-NEXT: [[TMP21:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP22:%.*]] = shl nuw nsw i64 [[TMP21]], 1
; CHECK-NEXT: [[MIN_EPILOG_ITERS_CHECK_NOT:%.*]] = icmp ugt i64 [[TMP6]], [[TMP22]]
; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK_NOT]], label [[SCALAR_PH]], label [[VEC_EPILOG_SCALAR_PH]]
; CHECK: vec.epilog.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC1]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT: [[TMP23:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP24:%.*]] = shl nuw nsw i64 [[TMP23]], 1
; CHECK-NEXT: [[TMP25:%.*]] = add nsw i64 [[TMP24]], -1
; CHECK-NEXT: [[N_MOD_VF4:%.*]] = and i64 [[N]], [[TMP25]]
; CHECK-NEXT: [[TMP26:%.*]] = icmp eq i64 [[N_MOD_VF4]], 0
; CHECK-NEXT: [[TMP27:%.*]] = select i1 [[TMP26]], i64 [[TMP24]], i64 [[N_MOD_VF4]]
; CHECK-NEXT: [[N_VEC5:%.*]] = sub i64 [[N]], [[TMP27]]
; CHECK-NEXT: [[TMP28:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP29:%.*]] = shl nuw nsw i64 [[TMP28]], 1
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[BC_RESUME_VAL]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP30:%.*]] = call <vscale x 2 x i64> @llvm.stepvector.nxv2i64()
; CHECK-NEXT: [[INDUCTION:%.*]] = add <vscale x 2 x i64> [[DOTSPLAT]], [[TMP30]]
; CHECK-NEXT: [[DOTSPLATINSERT8:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[TMP29]], i64 0
; CHECK-NEXT: [[DOTSPLAT9:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT8]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: vec.epilog.vector.body:
; CHECK-NEXT: [[INDEX7:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDEX_NEXT10:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 2 x i64> [ [[INDUCTION]], [[SCALAR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[TMP31:%.*]] = shl <vscale x 2 x i64> [[VEC_IND]], shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 1, i64 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[B]], <vscale x 2 x i64> [[TMP31]]
; CHECK-NEXT: [[WIDE_MASKED_GATHER1:%.*]] = call <vscale x 2 x float> @llvm.masked.gather.nxv2f32.nxv2p0(<vscale x 2 x ptr> [[TMP32]], i32 4, <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer), <vscale x 2 x float> poison)
; CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX7]]
; CHECK-NEXT: store <vscale x 2 x float> [[WIDE_MASKED_GATHER1]], ptr [[TMP33]], align 4
; CHECK-NEXT: [[INDEX_NEXT10]] = add nuw i64 [[INDEX7]], [[TMP29]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 2 x i64> [[VEC_IND]], [[DOTSPLAT9]]
; CHECK-NEXT: [[TMP34:%.*]] = icmp eq i64 [[INDEX_NEXT10]], [[N_VEC5]]
; CHECK-NEXT: br i1 [[TMP34]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: vec.epilog.middle.block:
; CHECK-NEXT: br label [[VEC_EPILOG_SCALAR_PH]]
; CHECK: vec.epilog.scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL6:%.*]] = phi i64 [ [[N_VEC5]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC1]], [[MIDDLE_BLOCK]] ], [ 0, [[ITER_CHECK:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY1:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY1]] ], [ [[BC_RESUME_VAL6]], [[VEC_EPILOG_SCALAR_PH]] ]
; CHECK-NEXT: [[ARRAYIDX_IDX:%.*]] = shl i64 [[INDVARS_IV]], 3
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[ARRAYIDX_IDX]]
; CHECK-NEXT: [[TMP16:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]]
; CHECK-NEXT: store float [[TMP16]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY1]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
;
Expand Down
Loading
Loading