shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
llvm-project/llvm/test/Transforms/LoopVectorize/if-pred-stores.ll
Florian Hahn 1de3dc7d23 [LV] Bail out early if BTC+1 wraps. 
Currently we fail to detect the case where BTC + 1 wraps, i.e. the
vector trip count is 0, In those cases, the minimum iteration count
check will fail, and the vector code will never be executed.

Explicitly check for this condition in computeMaxVF and avoid trying to
vectorize alltogether.

Note that a number of tests needed to be updated, because the vector
loop would never be executed given the input IR.

Fixes https://github.com/llvm/llvm-project/issues/122558.
2025-01-14 22:07:38 +00:00

695 lines
40 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S -vectorize-num-stores-pred=1 -force-vector-width=1 -force-vector-interleave=2 -passes=loop-vectorize,simplifycfg -verify-loop-info -simplifycfg-require-and-preserve-domtree=1 < %s | FileCheck %s --check-prefix=UNROLL
; RUN: opt -S -vectorize-num-stores-pred=1 -force-vector-width=1 -force-vector-interleave=2 -passes=loop-vectorize -verify-loop-info < %s | FileCheck %s --check-prefix=UNROLL-NOSIMPLIFY
; RUN: opt -S -vectorize-num-stores-pred=1 -force-vector-width=2 -force-vector-interleave=1 -passes=loop-vectorize,simplifycfg -verify-loop-info -simplifycfg-require-and-preserve-domtree=1 < %s | FileCheck %s --check-prefix=VEC
target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
; Test predication of stores.
define i32 @test(ptr nocapture %f) #0 {
; UNROLL-LABEL: @test(
; UNROLL-NEXT: entry:
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; UNROLL-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; UNROLL-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 1
; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[F:%.*]], i64 [[TMP0]]
; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[F]], i64 [[TMP1]]
; UNROLL-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP2]], align 4
; UNROLL-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP3]], align 4
; UNROLL-NEXT: [[TMP6:%.*]] = icmp sgt i32 [[TMP4]], 100
; UNROLL-NEXT: [[TMP7:%.*]] = icmp sgt i32 [[TMP5]], 100
; UNROLL-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; UNROLL: pred.store.if:
; UNROLL-NEXT: [[TMP8:%.*]] = add nsw i32 [[TMP4]], 20
; UNROLL-NEXT: store i32 [[TMP8]], ptr [[TMP2]], align 4
; UNROLL-NEXT: br label [[PRED_STORE_CONTINUE]]
; UNROLL: pred.store.continue:
; UNROLL-NEXT: br i1 [[TMP7]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; UNROLL: pred.store.if1:
; UNROLL-NEXT: [[TMP9:%.*]] = add nsw i32 [[TMP5]], 20
; UNROLL-NEXT: store i32 [[TMP9]], ptr [[TMP3]], align 4
; UNROLL-NEXT: br label [[PRED_STORE_CONTINUE2]]
; UNROLL: pred.store.continue2:
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; UNROLL-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 128
; UNROLL-NEXT: br i1 [[TMP10]], label [[FOR_END:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; UNROLL: for.end:
; UNROLL-NEXT: ret i32 0
;
; UNROLL-NOSIMPLIFY-LABEL: @test(
; UNROLL-NOSIMPLIFY-NEXT: entry:
; UNROLL-NOSIMPLIFY-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL-NOSIMPLIFY: vector.ph:
; UNROLL-NOSIMPLIFY-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL-NOSIMPLIFY: vector.body:
; UNROLL-NOSIMPLIFY-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; UNROLL-NOSIMPLIFY-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[F:%.*]], i64 [[TMP0]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[F]], i64 [[TMP1]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP2]], align 4
; UNROLL-NOSIMPLIFY-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP3]], align 4
; UNROLL-NOSIMPLIFY-NEXT: [[TMP6:%.*]] = icmp sgt i32 [[TMP4]], 100
; UNROLL-NOSIMPLIFY-NEXT: [[TMP7:%.*]] = icmp sgt i32 [[TMP5]], 100
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; UNROLL-NOSIMPLIFY: pred.store.if:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP8:%.*]] = add nsw i32 [[TMP4]], 20
; UNROLL-NOSIMPLIFY-NEXT: store i32 [[TMP8]], ptr [[TMP2]], align 4
; UNROLL-NOSIMPLIFY-NEXT: br label [[PRED_STORE_CONTINUE]]
; UNROLL-NOSIMPLIFY: pred.store.continue:
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[TMP7]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; UNROLL-NOSIMPLIFY: pred.store.if1:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP9:%.*]] = add nsw i32 [[TMP5]], 20
; UNROLL-NOSIMPLIFY-NEXT: store i32 [[TMP9]], ptr [[TMP3]], align 4
; UNROLL-NOSIMPLIFY-NEXT: br label [[PRED_STORE_CONTINUE2]]
; UNROLL-NOSIMPLIFY: pred.store.continue2:
; UNROLL-NOSIMPLIFY-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; UNROLL-NOSIMPLIFY-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 128
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; UNROLL-NOSIMPLIFY: middle.block:
; UNROLL-NOSIMPLIFY-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; UNROLL-NOSIMPLIFY: scalar.ph:
; UNROLL-NOSIMPLIFY-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 128, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_BODY:%.*]]
; UNROLL-NOSIMPLIFY: for.body:
; UNROLL-NOSIMPLIFY-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[F]], i64 [[INDVARS_IV]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP11:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; UNROLL-NOSIMPLIFY-NEXT: [[CMP1:%.*]] = icmp sgt i32 [[TMP11]], 100
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[CMP1]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; UNROLL-NOSIMPLIFY: if.then:
; UNROLL-NOSIMPLIFY-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], 20
; UNROLL-NOSIMPLIFY-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX]], align 4
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_INC]]
; UNROLL-NOSIMPLIFY: for.inc:
; UNROLL-NOSIMPLIFY-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 128
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[EXITCOND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; UNROLL-NOSIMPLIFY: for.end:
; UNROLL-NOSIMPLIFY-NEXT: ret i32 0
;
; VEC-LABEL: @test(
; VEC-NEXT: entry:
; VEC-NEXT: br label [[VECTOR_BODY:%.*]]
; VEC: vector.body:
; VEC-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; VEC-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; VEC-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[F:%.*]], i64 [[TMP0]]
; VEC-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
; VEC-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
; VEC-NEXT: [[TMP3:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD]], splat (i32 100)
; VEC-NEXT: [[TMP4:%.*]] = extractelement <2 x i1> [[TMP3]], i32 0
; VEC-NEXT: br i1 [[TMP4]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; VEC: pred.store.if:
; VEC-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[F]], i64 [[TMP0]]
; VEC-NEXT: [[TMP6:%.*]] = extractelement <2 x i32> [[WIDE_LOAD]], i32 0
; VEC-NEXT: [[TMP7:%.*]] = add nsw i32 [[TMP6]], 20
; VEC-NEXT: store i32 [[TMP7]], ptr [[TMP5]], align 4
; VEC-NEXT: br label [[PRED_STORE_CONTINUE]]
; VEC: pred.store.continue:
; VEC-NEXT: [[TMP8:%.*]] = extractelement <2 x i1> [[TMP3]], i32 1
; VEC-NEXT: br i1 [[TMP8]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; VEC: pred.store.if1:
; VEC-NEXT: [[TMP9:%.*]] = add i64 [[INDEX]], 1
; VEC-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[F]], i64 [[TMP9]]
; VEC-NEXT: [[TMP11:%.*]] = extractelement <2 x i32> [[WIDE_LOAD]], i32 1
; VEC-NEXT: [[TMP12:%.*]] = add nsw i32 [[TMP11]], 20
; VEC-NEXT: store i32 [[TMP12]], ptr [[TMP10]], align 4
; VEC-NEXT: br label [[PRED_STORE_CONTINUE2]]
; VEC: pred.store.continue2:
; VEC-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC-NEXT: [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 128
; VEC-NEXT: br i1 [[TMP13]], label [[FOR_END:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; VEC: for.end:
; VEC-NEXT: ret i32 0
;
entry:
br label %for.body
for.body:
%indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.inc ]
%arrayidx = getelementptr inbounds i32, ptr %f, i64 %indvars.iv
%0 = load i32, ptr %arrayidx, align 4
%cmp1 = icmp sgt i32 %0, 100
br i1 %cmp1, label %if.then, label %for.inc
if.then:
%add = add nsw i32 %0, 20
store i32 %add, ptr %arrayidx, align 4
br label %for.inc
for.inc:
%indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
%exitcond = icmp eq i64 %indvars.iv.next, 128
br i1 %exitcond, label %for.end, label %for.body
for.end:
ret i32 0
}
; Track basic blocks when unrolling conditional blocks. This code used to assert
; because we did not update the phi nodes with the proper predecessor in the
; vectorized loop body.
; PR18724
define void @bug18724(i1 %cond, ptr %ptr, i1 %cond.2, i64 %v.1, i32 %v.2) {
; UNROLL-LABEL: @bug18724(
; UNROLL-NEXT: entry:
; UNROLL-NEXT: [[TMP0:%.*]] = xor i1 [[COND:%.*]], true
; UNROLL-NEXT: call void @llvm.assume(i1 [[TMP0]])
; UNROLL-NEXT: [[TMP1:%.*]] = trunc i64 [[V_1:%.*]] to i32
; UNROLL-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP1]], i32 0)
; UNROLL-NEXT: [[TMP2:%.*]] = sub i32 [[SMAX]], [[TMP1]]
; UNROLL-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
; UNROLL-NEXT: [[TMP4:%.*]] = add nuw nsw i64 [[TMP3]], 1
; UNROLL-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP4]], 2
; UNROLL-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL: vector.ph:
; UNROLL-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP4]], 2
; UNROLL-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP4]], [[N_MOD_VF]]
; UNROLL-NEXT: [[IND_END:%.*]] = add i64 [[V_1]], [[N_VEC]]
; UNROLL-NEXT: [[TMP13:%.*]] = xor i1 [[COND_2:%.*]], true
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE3:%.*]] ]
; UNROLL-NEXT: [[VEC_PHI:%.*]] = phi i32 [ [[V_2:%.*]], [[VECTOR_PH]] ], [ [[PREDPHI:%.*]], [[PRED_STORE_CONTINUE3]] ]
; UNROLL-NEXT: [[VEC_PHI1:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[PREDPHI4:%.*]], [[PRED_STORE_CONTINUE3]] ]
; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[V_1]], [[INDEX]]
; UNROLL-NEXT: [[TMP5:%.*]] = add i64 [[OFFSET_IDX]], 0
; UNROLL-NEXT: [[TMP6:%.*]] = add i64 [[OFFSET_IDX]], 1
; UNROLL-NEXT: [[TMP7:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR:%.*]], i64 0, i64 [[TMP5]]
; UNROLL-NEXT: [[TMP8:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR]], i64 0, i64 [[TMP6]]
; UNROLL-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP7]], align 4
; UNROLL-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP8]], align 4
; UNROLL-NEXT: br i1 [[COND_2]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE3]]
; UNROLL: pred.store.if:
; UNROLL-NEXT: store i32 [[TMP9]], ptr [[TMP7]], align 4
; UNROLL-NEXT: store i32 [[TMP10]], ptr [[TMP8]], align 4
; UNROLL-NEXT: br label [[PRED_STORE_CONTINUE3]]
; UNROLL: pred.store.continue3:
; UNROLL-NEXT: [[TMP11:%.*]] = add i32 [[VEC_PHI]], 1
; UNROLL-NEXT: [[TMP12:%.*]] = add i32 [[VEC_PHI1]], 1
; UNROLL-NEXT: [[PREDPHI]] = select i1 [[TMP13]], i32 [[VEC_PHI]], i32 [[TMP11]]
; UNROLL-NEXT: [[PREDPHI4]] = select i1 [[TMP13]], i32 [[VEC_PHI1]], i32 [[TMP12]]
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; UNROLL-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; UNROLL-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; UNROLL: middle.block:
; UNROLL-NEXT: [[BIN_RDX:%.*]] = add i32 [[PREDPHI4]], [[PREDPHI]]
; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP4]], [[N_VEC]]
; UNROLL-NEXT: [[TMP16:%.*]] = xor i1 [[CMP_N]], true
; UNROLL-NEXT: call void @llvm.assume(i1 [[TMP16]])
; UNROLL-NEXT: br label [[SCALAR_PH]]
; UNROLL: scalar.ph:
; UNROLL-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[V_1]], [[ENTRY:%.*]] ]
; UNROLL-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ [[BIN_RDX]], [[MIDDLE_BLOCK]] ], [ [[V_2]], [[ENTRY]] ]
; UNROLL-NEXT: br label [[FOR_BODY14:%.*]]
; UNROLL: for.body14:
; UNROLL-NEXT: [[INDVARS_IV3:%.*]] = phi i64 [ [[INDVARS_IV_NEXT4:%.*]], [[FOR_INC23:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[INEWCHUNKS_120:%.*]] = phi i32 [ [[INEWCHUNKS_2:%.*]], [[FOR_INC23]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR]], i64 0, i64 [[INDVARS_IV3]]
; UNROLL-NEXT: [[TMP:%.*]] = load i32, ptr [[ARRAYIDX16]], align 4
; UNROLL-NEXT: br i1 [[COND_2]], label [[IF_THEN18:%.*]], label [[FOR_INC23]]
; UNROLL: if.then18:
; UNROLL-NEXT: store i32 [[TMP]], ptr [[ARRAYIDX16]], align 4
; UNROLL-NEXT: [[INC21:%.*]] = add nsw i32 [[INEWCHUNKS_120]], 1
; UNROLL-NEXT: br label [[FOR_INC23]]
; UNROLL: for.inc23:
; UNROLL-NEXT: [[INEWCHUNKS_2]] = phi i32 [ [[INC21]], [[IF_THEN18]] ], [ [[INEWCHUNKS_120]], [[FOR_BODY14]] ]
; UNROLL-NEXT: [[INDVARS_IV_NEXT4]] = add nsw i64 [[INDVARS_IV3]], 1
; UNROLL-NEXT: [[TMP1:%.*]] = trunc i64 [[INDVARS_IV3]] to i32
; UNROLL-NEXT: [[CMP13:%.*]] = icmp slt i32 [[TMP1]], 0
; UNROLL-NEXT: call void @llvm.assume(i1 [[CMP13]])
; UNROLL-NEXT: br label [[FOR_BODY14]]
;
; UNROLL-NOSIMPLIFY-LABEL: @bug18724(
; UNROLL-NOSIMPLIFY-NEXT: entry:
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_BODY9:%.*]]
; UNROLL-NOSIMPLIFY: for.body9:
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[COND:%.*]], label [[FOR_INC26:%.*]], label [[FOR_BODY14_PREHEADER:%.*]]
; UNROLL-NOSIMPLIFY: for.body14.preheader:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP0:%.*]] = trunc i64 [[V_1:%.*]] to i32
; UNROLL-NOSIMPLIFY-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP0]], i32 0)
; UNROLL-NOSIMPLIFY-NEXT: [[TMP1:%.*]] = sub i32 [[SMAX]], [[TMP0]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
; UNROLL-NOSIMPLIFY-NEXT: [[TMP3:%.*]] = add nuw nsw i64 [[TMP2]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP3]], 2
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL-NOSIMPLIFY: vector.ph:
; UNROLL-NOSIMPLIFY-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP3]], 2
; UNROLL-NOSIMPLIFY-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP3]], [[N_MOD_VF]]
; UNROLL-NOSIMPLIFY-NEXT: [[IND_END:%.*]] = add i64 [[V_1]], [[N_VEC]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP12:%.*]] = xor i1 [[COND_2:%.*]], true
; UNROLL-NOSIMPLIFY-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL-NOSIMPLIFY: vector.body:
; UNROLL-NOSIMPLIFY-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE3:%.*]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[VEC_PHI:%.*]] = phi i32 [ [[V_2:%.*]], [[VECTOR_PH]] ], [ [[PREDPHI:%.*]], [[PRED_STORE_CONTINUE3]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[VEC_PHI1:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[PREDPHI4:%.*]], [[PRED_STORE_CONTINUE3]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[V_1]], [[INDEX]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 0
; UNROLL-NOSIMPLIFY-NEXT: [[TMP5:%.*]] = add i64 [[OFFSET_IDX]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP6:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR:%.*]], i64 0, i64 [[TMP4]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP7:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR]], i64 0, i64 [[TMP5]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP6]], align 4
; UNROLL-NOSIMPLIFY-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP7]], align 4
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[COND_2]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; UNROLL-NOSIMPLIFY: pred.store.if:
; UNROLL-NOSIMPLIFY-NEXT: store i32 [[TMP8]], ptr [[TMP6]], align 4
; UNROLL-NOSIMPLIFY-NEXT: br label [[PRED_STORE_CONTINUE]]
; UNROLL-NOSIMPLIFY: pred.store.continue:
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[COND_2]], label [[PRED_STORE_IF2:%.*]], label [[PRED_STORE_CONTINUE3]]
; UNROLL-NOSIMPLIFY: pred.store.if2:
; UNROLL-NOSIMPLIFY-NEXT: store i32 [[TMP9]], ptr [[TMP7]], align 4
; UNROLL-NOSIMPLIFY-NEXT: br label [[PRED_STORE_CONTINUE3]]
; UNROLL-NOSIMPLIFY: pred.store.continue3:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP10:%.*]] = add i32 [[VEC_PHI]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP11:%.*]] = add i32 [[VEC_PHI1]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[PREDPHI]] = select i1 [[TMP12]], i32 [[VEC_PHI]], i32 [[TMP10]]
; UNROLL-NOSIMPLIFY-NEXT: [[PREDPHI4]] = select i1 [[TMP12]], i32 [[VEC_PHI1]], i32 [[TMP11]]
; UNROLL-NOSIMPLIFY-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; UNROLL-NOSIMPLIFY-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; UNROLL-NOSIMPLIFY: middle.block:
; UNROLL-NOSIMPLIFY-NEXT: [[BIN_RDX:%.*]] = add i32 [[PREDPHI4]], [[PREDPHI]]
; UNROLL-NOSIMPLIFY-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP3]], [[N_VEC]]
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[CMP_N]], label [[FOR_INC26_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; UNROLL-NOSIMPLIFY: scalar.ph:
; UNROLL-NOSIMPLIFY-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[V_1]], [[FOR_BODY14_PREHEADER]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ [[BIN_RDX]], [[MIDDLE_BLOCK]] ], [ [[V_2]], [[FOR_BODY14_PREHEADER]] ]
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_BODY14:%.*]]
; UNROLL-NOSIMPLIFY: for.body14:
; UNROLL-NOSIMPLIFY-NEXT: [[INDVARS_IV3:%.*]] = phi i64 [ [[INDVARS_IV_NEXT4:%.*]], [[FOR_INC23:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[INEWCHUNKS_120:%.*]] = phi i32 [ [[INEWCHUNKS_2:%.*]], [[FOR_INC23]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR]], i64 0, i64 [[INDVARS_IV3]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP:%.*]] = load i32, ptr [[ARRAYIDX16]], align 4
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[COND_2]], label [[IF_THEN18:%.*]], label [[FOR_INC23]]
; UNROLL-NOSIMPLIFY: if.then18:
; UNROLL-NOSIMPLIFY-NEXT: store i32 [[TMP]], ptr [[ARRAYIDX16]], align 4
; UNROLL-NOSIMPLIFY-NEXT: [[INC21:%.*]] = add nsw i32 [[INEWCHUNKS_120]], 1
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_INC23]]
; UNROLL-NOSIMPLIFY: for.inc23:
; UNROLL-NOSIMPLIFY-NEXT: [[INEWCHUNKS_2]] = phi i32 [ [[INC21]], [[IF_THEN18]] ], [ [[INEWCHUNKS_120]], [[FOR_BODY14]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[INDVARS_IV_NEXT4]] = add nsw i64 [[INDVARS_IV3]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP1:%.*]] = trunc i64 [[INDVARS_IV3]] to i32
; UNROLL-NOSIMPLIFY-NEXT: [[CMP13:%.*]] = icmp slt i32 [[TMP1]], 0
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[CMP13]], label [[FOR_BODY14]], label [[FOR_INC26_LOOPEXIT]], !llvm.loop [[LOOP5:![0-9]+]]
; UNROLL-NOSIMPLIFY: for.inc26.loopexit:
; UNROLL-NOSIMPLIFY-NEXT: [[INEWCHUNKS_2_LCSSA:%.*]] = phi i32 [ [[INEWCHUNKS_2]], [[FOR_INC23]] ], [ [[BIN_RDX]], [[MIDDLE_BLOCK]] ]
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_INC26]]
; UNROLL-NOSIMPLIFY: for.inc26:
; UNROLL-NOSIMPLIFY-NEXT: [[INEWCHUNKS_1_LCSSA:%.*]] = phi i32 [ undef, [[FOR_BODY9]] ], [ [[INEWCHUNKS_2_LCSSA]], [[FOR_INC26_LOOPEXIT]] ]
; UNROLL-NOSIMPLIFY-NEXT: unreachable
;
; VEC-LABEL: @bug18724(
; VEC-NEXT: entry:
; VEC-NEXT: [[TMP0:%.*]] = xor i1 [[COND:%.*]], true
; VEC-NEXT: call void @llvm.assume(i1 [[TMP0]])
; VEC-NEXT: [[TMP1:%.*]] = trunc i64 [[V_1:%.*]] to i32
; VEC-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP1]], i32 0)
; VEC-NEXT: [[TMP2:%.*]] = sub i32 [[SMAX]], [[TMP1]]
; VEC-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
; VEC-NEXT: [[TMP4:%.*]] = add nuw nsw i64 [[TMP3]], 1
; VEC-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP4]], 2
; VEC-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VEC: vector.ph:
; VEC-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP4]], 2
; VEC-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP4]], [[N_MOD_VF]]
; VEC-NEXT: [[IND_END:%.*]] = add i64 [[V_1]], [[N_VEC]]
; VEC-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i1> poison, i1 [[COND_2:%.*]], i64 0
; VEC-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i1> [[BROADCAST_SPLATINSERT]], <2 x i1> poison, <2 x i32> zeroinitializer
; VEC-NEXT: [[TMP17:%.*]] = xor <2 x i1> [[BROADCAST_SPLAT]], splat (i1 true)
; VEC-NEXT: [[TMP5:%.*]] = insertelement <2 x i32> zeroinitializer, i32 [[V_2:%.*]], i32 0
; VEC-NEXT: br label [[VECTOR_BODY:%.*]]
; VEC: vector.body:
; VEC-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; VEC-NEXT: [[VEC_PHI:%.*]] = phi <2 x i32> [ [[TMP5]], [[VECTOR_PH]] ], [ [[PREDPHI:%.*]], [[PRED_STORE_CONTINUE2]] ]
; VEC-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[V_1]], [[INDEX]]
; VEC-NEXT: [[TMP6:%.*]] = add i64 [[OFFSET_IDX]], 0
; VEC-NEXT: [[TMP7:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR:%.*]], i64 0, i64 [[TMP6]]
; VEC-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP7]], i32 0
; VEC-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP8]], align 4
; VEC-NEXT: [[TMP9:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 0
; VEC-NEXT: br i1 [[TMP9]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; VEC: pred.store.if:
; VEC-NEXT: [[TMP10:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR]], i64 0, i64 [[TMP6]]
; VEC-NEXT: [[TMP11:%.*]] = extractelement <2 x i32> [[WIDE_LOAD]], i32 0
; VEC-NEXT: store i32 [[TMP11]], ptr [[TMP10]], align 4
; VEC-NEXT: br label [[PRED_STORE_CONTINUE]]
; VEC: pred.store.continue:
; VEC-NEXT: [[TMP12:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 1
; VEC-NEXT: br i1 [[TMP12]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; VEC: pred.store.if1:
; VEC-NEXT: [[TMP13:%.*]] = add i64 [[OFFSET_IDX]], 1
; VEC-NEXT: [[TMP14:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR]], i64 0, i64 [[TMP13]]
; VEC-NEXT: [[TMP15:%.*]] = extractelement <2 x i32> [[WIDE_LOAD]], i32 1
; VEC-NEXT: store i32 [[TMP15]], ptr [[TMP14]], align 4
; VEC-NEXT: br label [[PRED_STORE_CONTINUE2]]
; VEC: pred.store.continue2:
; VEC-NEXT: [[TMP16:%.*]] = add <2 x i32> [[VEC_PHI]], splat (i32 1)
; VEC-NEXT: [[PREDPHI]] = select <2 x i1> [[TMP17]], <2 x i32> [[VEC_PHI]], <2 x i32> [[TMP16]]
; VEC-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VEC-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; VEC: middle.block:
; VEC-NEXT: [[TMP19:%.*]] = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> [[PREDPHI]])
; VEC-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP4]], [[N_VEC]]
; VEC-NEXT: [[TMP20:%.*]] = xor i1 [[CMP_N]], true
; VEC-NEXT: call void @llvm.assume(i1 [[TMP20]])
; VEC-NEXT: br label [[SCALAR_PH]]
; VEC: scalar.ph:
; VEC-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[V_1]], [[ENTRY:%.*]] ]
; VEC-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ [[TMP19]], [[MIDDLE_BLOCK]] ], [ [[V_2]], [[ENTRY]] ]
; VEC-NEXT: br label [[FOR_BODY14:%.*]]
; VEC: for.body14:
; VEC-NEXT: [[INDVARS_IV3:%.*]] = phi i64 [ [[INDVARS_IV_NEXT4:%.*]], [[FOR_INC23:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC-NEXT: [[INEWCHUNKS_120:%.*]] = phi i32 [ [[INEWCHUNKS_2:%.*]], [[FOR_INC23]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; VEC-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds [768 x i32], ptr [[PTR]], i64 0, i64 [[INDVARS_IV3]]
; VEC-NEXT: [[TMP:%.*]] = load i32, ptr [[ARRAYIDX16]], align 4
; VEC-NEXT: br i1 [[COND_2]], label [[IF_THEN18:%.*]], label [[FOR_INC23]]
; VEC: if.then18:
; VEC-NEXT: store i32 [[TMP]], ptr [[ARRAYIDX16]], align 4
; VEC-NEXT: [[INC21:%.*]] = add nsw i32 [[INEWCHUNKS_120]], 1
; VEC-NEXT: br label [[FOR_INC23]]
; VEC: for.inc23:
; VEC-NEXT: [[INEWCHUNKS_2]] = phi i32 [ [[INC21]], [[IF_THEN18]] ], [ [[INEWCHUNKS_120]], [[FOR_BODY14]] ]
; VEC-NEXT: [[INDVARS_IV_NEXT4]] = add nsw i64 [[INDVARS_IV3]], 1
; VEC-NEXT: [[TMP1:%.*]] = trunc i64 [[INDVARS_IV3]] to i32
; VEC-NEXT: [[CMP13:%.*]] = icmp slt i32 [[TMP1]], 0
; VEC-NEXT: call void @llvm.assume(i1 [[CMP13]])
; VEC-NEXT: br label [[FOR_BODY14]]
;
entry:
br label %for.body9
for.body9:
br i1 %cond, label %for.inc26, label %for.body14
for.body14:
%indvars.iv3 = phi i64 [ %indvars.iv.next4, %for.inc23 ], [ %v.1, %for.body9 ]
%iNewChunks.120 = phi i32 [ %iNewChunks.2, %for.inc23 ], [ %v.2, %for.body9 ]
%arrayidx16 = getelementptr inbounds [768 x i32], ptr %ptr, i64 0, i64 %indvars.iv3
%tmp = load i32, ptr %arrayidx16, align 4
br i1 %cond.2, label %if.then18, label %for.inc23
if.then18:
store i32 %tmp, ptr %arrayidx16, align 4
%inc21 = add nsw i32 %iNewChunks.120, 1
br label %for.inc23
for.inc23:
%iNewChunks.2 = phi i32 [ %inc21, %if.then18 ], [ %iNewChunks.120, %for.body14 ]
%indvars.iv.next4 = add nsw i64 %indvars.iv3, 1
%tmp1 = trunc i64 %indvars.iv3 to i32
%cmp13 = icmp slt i32 %tmp1, 0
br i1 %cmp13, label %for.body14, label %for.inc26
for.inc26:
%iNewChunks.1.lcssa = phi i32 [ undef, %for.body9 ], [ %iNewChunks.2, %for.inc23 ]
unreachable
}
; In the test below, it's more profitable for the expression feeding the
; conditional store to remain scalar. Since we can only type-shrink vector
; types, we shouldn't try to represent the expression in a smaller type.
;
define void @minimal_bit_widths(i1 %c) {
; UNROLL-LABEL: @minimal_bit_widths(
; UNROLL-NEXT: entry:
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; UNROLL-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; UNROLL-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 1
; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr undef, i64 [[TMP0]]
; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr undef, i64 [[TMP1]]
; UNROLL-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP2]], align 1
; UNROLL-NEXT: [[TMP5:%.*]] = load i8, ptr [[TMP3]], align 1
; UNROLL-NEXT: br i1 [[C:%.*]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE2]]
; UNROLL: pred.store.if:
; UNROLL-NEXT: store i8 [[TMP4]], ptr [[TMP2]], align 1
; UNROLL-NEXT: store i8 [[TMP5]], ptr [[TMP3]], align 1
; UNROLL-NEXT: br label [[PRED_STORE_CONTINUE2]]
; UNROLL: pred.store.continue2:
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; UNROLL-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; UNROLL-NEXT: br i1 [[TMP6]], label [[FOR_END:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; UNROLL: for.end:
; UNROLL-NEXT: ret void
;
; UNROLL-NOSIMPLIFY-LABEL: @minimal_bit_widths(
; UNROLL-NOSIMPLIFY-NEXT: entry:
; UNROLL-NOSIMPLIFY-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL-NOSIMPLIFY: vector.ph:
; UNROLL-NOSIMPLIFY-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL-NOSIMPLIFY: vector.body:
; UNROLL-NOSIMPLIFY-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; UNROLL-NOSIMPLIFY-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr undef, i64 [[TMP0]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr undef, i64 [[TMP1]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP5:%.*]] = load i8, ptr [[TMP3]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[C:%.*]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; UNROLL-NOSIMPLIFY: pred.store.if:
; UNROLL-NOSIMPLIFY-NEXT: store i8 [[TMP4]], ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br label [[PRED_STORE_CONTINUE]]
; UNROLL-NOSIMPLIFY: pred.store.continue:
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[C]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; UNROLL-NOSIMPLIFY: pred.store.if1:
; UNROLL-NOSIMPLIFY-NEXT: store i8 [[TMP5]], ptr [[TMP3]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br label [[PRED_STORE_CONTINUE2]]
; UNROLL-NOSIMPLIFY: pred.store.continue2:
; UNROLL-NOSIMPLIFY-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; UNROLL-NOSIMPLIFY-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; UNROLL-NOSIMPLIFY: middle.block:
; UNROLL-NOSIMPLIFY-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; UNROLL-NOSIMPLIFY: scalar.ph:
; UNROLL-NOSIMPLIFY-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i64 [ 0, [[MIDDLE_BLOCK]] ], [ 1000, [[ENTRY]] ]
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_BODY:%.*]]
; UNROLL-NOSIMPLIFY: for.body:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP0:%.*]] = phi i64 [ [[TMP6:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP1:%.*]] = phi i64 [ [[TMP7:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr undef, i64 [[TMP0]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP3:%.*]] = load i8, ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; UNROLL-NOSIMPLIFY: if.then:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP4:%.*]] = zext i8 [[TMP3]] to i32
; UNROLL-NOSIMPLIFY-NEXT: [[TMP5:%.*]] = trunc i32 [[TMP4]] to i8
; UNROLL-NOSIMPLIFY-NEXT: store i8 [[TMP5]], ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_INC]]
; UNROLL-NOSIMPLIFY: for.inc:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP6]] = add nuw nsw i64 [[TMP0]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP7]] = add i64 [[TMP1]], -1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP8:%.*]] = icmp eq i64 [[TMP7]], 0
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[TMP8]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; UNROLL-NOSIMPLIFY: for.end:
; UNROLL-NOSIMPLIFY-NEXT: ret void
;
; VEC-LABEL: @minimal_bit_widths(
; VEC-NEXT: entry:
; VEC-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i1> poison, i1 [[C:%.*]], i64 0
; VEC-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i1> [[BROADCAST_SPLATINSERT]], <2 x i1> poison, <2 x i32> zeroinitializer
; VEC-NEXT: br label [[VECTOR_BODY:%.*]]
; VEC: vector.body:
; VEC-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; VEC-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; VEC-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr undef, i64 [[TMP0]]
; VEC-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP1]], i32 0
; VEC-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i8>, ptr [[TMP2]], align 1
; VEC-NEXT: [[TMP3:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 0
; VEC-NEXT: br i1 [[TMP3]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; VEC: pred.store.if:
; VEC-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr undef, i64 [[TMP0]]
; VEC-NEXT: [[TMP5:%.*]] = extractelement <2 x i8> [[WIDE_LOAD]], i32 0
; VEC-NEXT: store i8 [[TMP5]], ptr [[TMP4]], align 1
; VEC-NEXT: br label [[PRED_STORE_CONTINUE]]
; VEC: pred.store.continue:
; VEC-NEXT: [[TMP6:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 1
; VEC-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; VEC: pred.store.if1:
; VEC-NEXT: [[TMP7:%.*]] = add i64 [[INDEX]], 1
; VEC-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr undef, i64 [[TMP7]]
; VEC-NEXT: [[TMP9:%.*]] = extractelement <2 x i8> [[WIDE_LOAD]], i32 1
; VEC-NEXT: store i8 [[TMP9]], ptr [[TMP8]], align 1
; VEC-NEXT: br label [[PRED_STORE_CONTINUE2]]
; VEC: pred.store.continue2:
; VEC-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; VEC-NEXT: br i1 [[TMP10]], label [[FOR_END:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; VEC: for.end:
; VEC-NEXT: ret void
;
entry:
br label %for.body
for.body:
%tmp0 = phi i64 [ %tmp6, %for.inc ], [ 0, %entry ]
%tmp1 = phi i64 [ %tmp7, %for.inc ], [ 1000, %entry ]
%tmp2 = getelementptr i8, ptr undef, i64 %tmp0
%tmp3 = load i8, ptr %tmp2, align 1
br i1 %c, label %if.then, label %for.inc
if.then:
%tmp4 = zext i8 %tmp3 to i32
%tmp5 = trunc i32 %tmp4 to i8
store i8 %tmp5, ptr %tmp2, align 1
br label %for.inc
for.inc:
%tmp6 = add nuw nsw i64 %tmp0, 1
%tmp7 = add i64 %tmp1, -1
%tmp8 = icmp eq i64 %tmp7, 0
br i1 %tmp8, label %for.end, label %for.body
for.end:
ret void
}
define void @minimal_bit_widths_with_aliasing_store(i1 %c, ptr %ptr) {
; UNROLL-LABEL: @minimal_bit_widths_with_aliasing_store(
; UNROLL-NEXT: entry:
; UNROLL-NEXT: br label [[FOR_BODY:%.*]]
; UNROLL: vector.body:
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[FOR_INC:%.*]] ]
; UNROLL-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; UNROLL-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 1
; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP0]]
; UNROLL-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP1]]
; UNROLL-NEXT: [[TMP3:%.*]] = load i8, ptr [[TMP2]], align 1
; UNROLL-NEXT: [[TMP5:%.*]] = load i8, ptr [[TMP4]], align 1
; UNROLL-NEXT: store i8 0, ptr [[TMP2]], align 1
; UNROLL-NEXT: store i8 0, ptr [[TMP4]], align 1
; UNROLL-NEXT: br i1 [[C:%.*]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; UNROLL: pred.store.if:
; UNROLL-NEXT: store i8 [[TMP3]], ptr [[TMP2]], align 1
; UNROLL-NEXT: store i8 [[TMP5]], ptr [[TMP4]], align 1
; UNROLL-NEXT: br label [[FOR_INC]]
; UNROLL: pred.store.continue2:
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; UNROLL-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; UNROLL-NEXT: br i1 [[TMP6]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; UNROLL: for.end:
; UNROLL-NEXT: ret void
;
; UNROLL-NOSIMPLIFY-LABEL: @minimal_bit_widths_with_aliasing_store(
; UNROLL-NOSIMPLIFY-NEXT: entry:
; UNROLL-NOSIMPLIFY-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL-NOSIMPLIFY: vector.ph:
; UNROLL-NOSIMPLIFY-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL-NOSIMPLIFY: vector.body:
; UNROLL-NOSIMPLIFY-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; UNROLL-NOSIMPLIFY-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP0]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP1]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP5:%.*]] = load i8, ptr [[TMP3]], align 1
; UNROLL-NOSIMPLIFY-NEXT: store i8 0, ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: store i8 0, ptr [[TMP3]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[C:%.*]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; UNROLL-NOSIMPLIFY: pred.store.if:
; UNROLL-NOSIMPLIFY-NEXT: store i8 [[TMP4]], ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br label [[PRED_STORE_CONTINUE]]
; UNROLL-NOSIMPLIFY: pred.store.continue:
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[C]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; UNROLL-NOSIMPLIFY: pred.store.if1:
; UNROLL-NOSIMPLIFY-NEXT: store i8 [[TMP5]], ptr [[TMP3]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br label [[PRED_STORE_CONTINUE2]]
; UNROLL-NOSIMPLIFY: pred.store.continue2:
; UNROLL-NOSIMPLIFY-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; UNROLL-NOSIMPLIFY-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; UNROLL-NOSIMPLIFY: middle.block:
; UNROLL-NOSIMPLIFY-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; UNROLL-NOSIMPLIFY: scalar.ph:
; UNROLL-NOSIMPLIFY-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i64 [ 0, [[MIDDLE_BLOCK]] ], [ 1000, [[ENTRY]] ]
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_BODY:%.*]]
; UNROLL-NOSIMPLIFY: for.body:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP0:%.*]] = phi i64 [ [[TMP6:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP1:%.*]] = phi i64 [ [[TMP7:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP0]]
; UNROLL-NOSIMPLIFY-NEXT: [[TMP3:%.*]] = load i8, ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: store i8 0, ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; UNROLL-NOSIMPLIFY: if.then:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP4:%.*]] = zext i8 [[TMP3]] to i32
; UNROLL-NOSIMPLIFY-NEXT: [[TMP5:%.*]] = trunc i32 [[TMP4]] to i8
; UNROLL-NOSIMPLIFY-NEXT: store i8 [[TMP5]], ptr [[TMP2]], align 1
; UNROLL-NOSIMPLIFY-NEXT: br label [[FOR_INC]]
; UNROLL-NOSIMPLIFY: for.inc:
; UNROLL-NOSIMPLIFY-NEXT: [[TMP6]] = add nuw nsw i64 [[TMP0]], 1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP7]] = add i64 [[TMP1]], -1
; UNROLL-NOSIMPLIFY-NEXT: [[TMP8:%.*]] = icmp eq i64 [[TMP7]], 0
; UNROLL-NOSIMPLIFY-NEXT: br i1 [[TMP8]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; UNROLL-NOSIMPLIFY: for.end:
; UNROLL-NOSIMPLIFY-NEXT: ret void
;
; VEC-LABEL: @minimal_bit_widths_with_aliasing_store(
; VEC-NEXT: entry:
; VEC-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i1> poison, i1 [[C1:%.*]], i64 0
; VEC-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i1> [[BROADCAST_SPLATINSERT]], <2 x i1> poison, <2 x i32> zeroinitializer
; VEC-NEXT: br label [[FOR_BODY:%.*]]
; VEC: vector.body:
; VEC-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; VEC-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; VEC-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP0]]
; VEC-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP2]], i32 0
; VEC-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i8>, ptr [[TMP3]], align 1
; VEC-NEXT: store <2 x i8> zeroinitializer, ptr [[TMP3]], align 1
; VEC-NEXT: [[C:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 0
; VEC-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[FOR_INC:%.*]]
; VEC: pred.store.if:
; VEC-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP0]]
; VEC-NEXT: [[TMP5:%.*]] = extractelement <2 x i8> [[WIDE_LOAD]], i32 0
; VEC-NEXT: store i8 [[TMP5]], ptr [[TMP4]], align 1
; VEC-NEXT: br label [[FOR_INC]]
; VEC: pred.store.continue:
; VEC-NEXT: [[TMP6:%.*]] = extractelement <2 x i1> [[BROADCAST_SPLAT]], i32 1
; VEC-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; VEC: pred.store.if1:
; VEC-NEXT: [[TMP7:%.*]] = add i64 [[INDEX]], 1
; VEC-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP7]]
; VEC-NEXT: [[TMP9:%.*]] = extractelement <2 x i8> [[WIDE_LOAD]], i32 1
; VEC-NEXT: store i8 [[TMP9]], ptr [[TMP8]], align 1
; VEC-NEXT: br label [[PRED_STORE_CONTINUE2]]
; VEC: pred.store.continue2:
; VEC-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; VEC-NEXT: br i1 [[TMP10]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; VEC: for.end:
; VEC-NEXT: ret void
;
entry:
br label %for.body
for.body:
%tmp0 = phi i64 [ %tmp6, %for.inc ], [ 0, %entry ]
%tmp1 = phi i64 [ %tmp7, %for.inc ], [ 1000, %entry ]
%tmp2 = getelementptr i8, ptr %ptr, i64 %tmp0
%tmp3 = load i8, ptr %tmp2, align 1
store i8 0, ptr %tmp2
br i1 %c, label %if.then, label %for.inc
if.then:
%tmp4 = zext i8 %tmp3 to i32
%tmp5 = trunc i32 %tmp4 to i8
store i8 %tmp5, ptr %tmp2, align 1
br label %for.inc
for.inc:
%tmp6 = add nuw nsw i64 %tmp0, 1
%tmp7 = add i64 %tmp1, -1
%tmp8 = icmp eq i64 %tmp7, 0
br i1 %tmp8, label %for.end, label %for.body
for.end:
ret void
}
Reference in New Issue View Git Blame Copy Permalink