[llvm] 5c7ae10 - [LV] Add store to test to make sure the loop is not dead.
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Sun Feb 20 07:06:10 PST 2022
Author: Florian Hahn
Date: 2022-02-20T15:05:29Z
New Revision: 5c7ae10ceca09832efd5a4161cf468368665a51b
URL: https://github.com/llvm/llvm-project/commit/5c7ae10ceca09832efd5a4161cf468368665a51b
DIFF: https://github.com/llvm/llvm-project/commit/5c7ae10ceca09832efd5a4161cf468368665a51b.diff
LOG: [LV] Add store to test to make sure the loop is not dead.
Add an extra store to the test, to make sure the operations in the loop
cannot be optimized away after D118051.
Added:
Modified:
llvm/test/Transforms/LoopVectorize/first-order-recurrence-complex.ll
Removed:
################################################################################
diff --git a/llvm/test/Transforms/LoopVectorize/first-order-recurrence-complex.ll b/llvm/test/Transforms/LoopVectorize/first-order-recurrence-complex.ll
index 70b8c81737e7..e7393b85b552 100644
--- a/llvm/test/Transforms/LoopVectorize/first-order-recurrence-complex.ll
+++ b/llvm/test/Transforms/LoopVectorize/first-order-recurrence-complex.ll
@@ -374,73 +374,74 @@ exit:
}
; Sink %tmp38 after %tmp60, then it enable the loop vectorization.
-define void @instruction_with_2_FOR_operands() {
+define void @instruction_with_2_FOR_operands(float* noalias %A, float* noalias %B, float* noalias %C) {
; CHECK-LABEL: @instruction_with_2_FOR_operands(
; CHECK-NEXT: bb:
-; CHECK-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 undef, i64 0)
-; CHECK-NEXT: [[TMP0:%.*]] = add nuw i64 [[SMAX]], 1
-; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 4
-; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 4
-; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VECTOR_RECUR:%.*]] = phi <4 x float> [ <float poison, float poison, float poison, float undef>, [[VECTOR_PH]] ], [ [[BROADCAST_SPLAT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VECTOR_RECUR1:%.*]] = phi <4 x float> [ <float poison, float poison, float poison, float undef>, [[VECTOR_PH]] ], [ [[BROADCAST_SPLAT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT: [[TMP2:%.*]] = load float, float* undef, align 4
+; CHECK-NEXT: [[VECTOR_RECUR:%.*]] = phi <4 x float> [ <float poison, float poison, float poison, float 0.000000e+00>, [[VECTOR_PH]] ], [ [[BROADCAST_SPLAT3:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VECTOR_RECUR1:%.*]] = phi <4 x float> [ <float poison, float poison, float poison, float 1.000000e+00>, [[VECTOR_PH]] ], [ [[BROADCAST_SPLAT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, float* [[C:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP2:%.*]] = load float, float* [[A:%.*]], align 4
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x float> poison, float [[TMP2]], i32 0
; CHECK-NEXT: [[BROADCAST_SPLAT]] = shufflevector <4 x float> [[BROADCAST_SPLATINSERT]], <4 x float> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <4 x float> [[VECTOR_RECUR1]], <4 x float> [[BROADCAST_SPLAT]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT: [[TMP4:%.*]] = load float, float* undef, align 4
+; CHECK-NEXT: [[TMP4:%.*]] = load float, float* [[B:%.*]], align 4
; CHECK-NEXT: [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <4 x float> poison, float [[TMP4]], i32 0
; CHECK-NEXT: [[BROADCAST_SPLAT3]] = shufflevector <4 x float> [[BROADCAST_SPLATINSERT2]], <4 x float> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <4 x float> [[VECTOR_RECUR]], <4 x float> [[BROADCAST_SPLAT3]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
; CHECK-NEXT: [[TMP6:%.*]] = fmul fast <4 x float> [[TMP5]], [[TMP3]]
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, float* [[TMP1]], i32 0
+; CHECK-NEXT: [[TMP8:%.*]] = bitcast float* [[TMP7]] to <4 x float>*
+; CHECK-NEXT: store <4 x float> [[TMP6]], <4 x float>* [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: middle.block:
-; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1001, 1000
; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x float> [[BROADCAST_SPLAT3]], i32 3
; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT_FOR_PHI:%.*]] = extractelement <4 x float> [[BROADCAST_SPLAT3]], i32 2
; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT4:%.*]] = extractelement <4 x float> [[BROADCAST_SPLAT]], i32 3
; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT_FOR_PHI5:%.*]] = extractelement <4 x float> [[BROADCAST_SPLAT]], i32 2
; CHECK-NEXT: br i1 [[CMP_N]], label [[BB74:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
-; CHECK-NEXT: [[SCALAR_RECUR_INIT6:%.*]] = phi float [ undef, [[BB:%.*]] ], [ [[VECTOR_RECUR_EXTRACT4]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT: [[SCALAR_RECUR_INIT:%.*]] = phi float [ undef, [[BB]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[BB]] ]
+; CHECK-NEXT: [[SCALAR_RECUR_INIT6:%.*]] = phi float [ 1.000000e+00, [[BB:%.*]] ], [ [[VECTOR_RECUR_EXTRACT4]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: [[SCALAR_RECUR_INIT:%.*]] = phi float [ 0.000000e+00, [[BB]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[BB]] ]
; CHECK-NEXT: br label [[BB13:%.*]]
; CHECK: bb13:
; CHECK-NEXT: [[SCALAR_RECUR:%.*]] = phi float [ [[TMP60:%.*]], [[BB13]] ], [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[SCALAR_RECUR7:%.*]] = phi float [ [[TMP49:%.*]], [[BB13]] ], [ [[SCALAR_RECUR_INIT6]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[BB13]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
+; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[BB13]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[TMP38:%.*]] = fmul fast float [[SCALAR_RECUR]], [[SCALAR_RECUR7]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[TMP49]] = load float, float* undef, align 4
-; CHECK-NEXT: [[TMP60]] = load float, float* undef, align 4
-; CHECK-NEXT: [[TMP12:%.*]] = icmp slt i64 [[INDVARS_IV]], undef
+; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds float, float* [[C]], i64 [[IV]]
+; CHECK-NEXT: [[TMP49]] = load float, float* [[A]], align 4
+; CHECK-NEXT: [[TMP60]] = load float, float* [[B]], align 4
+; CHECK-NEXT: store float [[TMP38]], float* [[GEP]], align 4
+; CHECK-NEXT: [[TMP12:%.*]] = icmp slt i64 [[IV]], 1000
; CHECK-NEXT: br i1 [[TMP12]], label [[BB13]], label [[BB74]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK: bb74:
; CHECK-NEXT: ret void
;
-
-
bb:
br label %bb13
bb13: ; preds = %bb13, %bb
- %tmp37 = phi float [ %tmp60, %bb13 ], [ undef, %bb ]
- %tmp27 = phi float [ %tmp49, %bb13 ], [ undef, %bb ]
- %indvars.iv = phi i64 [ %indvars.iv.next, %bb13 ], [ 0, %bb ]
+ %tmp37 = phi float [ %tmp60, %bb13 ], [ 0.0, %bb ]
+ %tmp27 = phi float [ %tmp49, %bb13 ], [ 1.0, %bb ]
+ %iv = phi i64 [ %iv.next, %bb13 ], [ 0, %bb ]
%tmp38 = fmul fast float %tmp37, %tmp27
- %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
- %tmp49 = load float, float* undef, align 4
- %tmp60 = load float, float* undef, align 4
- %tmp12 = icmp slt i64 %indvars.iv, undef
+ %iv.next = add nuw nsw i64 %iv, 1
+ %gep = getelementptr inbounds float, float* %C, i64 %iv
+ %tmp49 = load float, float* %A, align 4
+ %tmp60 = load float, float* %B, align 4
+ store float %tmp38, float* %gep
+ %tmp12 = icmp slt i64 %iv, 1000
br i1 %tmp12, label %bb13, label %bb74
bb74: ; preds = %bb13
More information about the llvm-commits
mailing list