From 97469d4c20615e7e7986a23d6c42a59f011590a3 Mon Sep 17 00:00:00 2001 From: Florian Hahn Date: Thu, 5 Aug 2021 10:33:29 +0100 Subject: [PATCH] [SLP] Add additional memory version tests. --- .../SLPVectorizer/AArch64/memory-runtime-checks.ll | 100 +++++++++++++++++++++ 1 file changed, 100 insertions(+) diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/memory-runtime-checks.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/memory-runtime-checks.ll index 3ecb0f2..dc7e2a1 100644 --- a/llvm/test/Transforms/SLPVectorizer/AArch64/memory-runtime-checks.ll +++ b/llvm/test/Transforms/SLPVectorizer/AArch64/memory-runtime-checks.ll @@ -352,3 +352,103 @@ bb1: ; preds = %bb bb15: ; preds = %bb15, %bb14 br label %bb15 } + +%struct = type { i32, i32, float, float } + +; Some points we collected as candidates for runtime checks have been removed +; before generating runtime checks. Make sure versioning is skipped. +define void @test_bounds_removed_before_runtime_checks(%struct * %A, i32** %B, i1 %c) { +; CHECK-LABEL: @test_bounds_removed_before_runtime_checks( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT:%.*]], %struct* [[A:%.*]], i64 0, i32 0 +; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT]], %struct* [[A]], i64 0, i32 1 +; CHECK-NEXT: [[TMP0:%.*]] = bitcast i32* [[TMP11]] to <2 x i32>* +; CHECK-NEXT: store <2 x i32> , <2 x i32>* [[TMP0]], align 8 +; CHECK-NEXT: [[TMP13:%.*]] = load i32*, i32** [[B:%.*]], align 8 +; CHECK-NEXT: br i1 [[C:%.*]], label [[BB23:%.*]], label [[BB14:%.*]] +; CHECK: bb14: +; CHECK-NEXT: [[TMP15:%.*]] = sext i32 10 to i64 +; CHECK-NEXT: [[TMP16:%.*]] = add nsw i64 2, [[TMP15]] +; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, i32* [[TMP13]], i64 [[TMP16]] +; CHECK-NEXT: [[TMP18:%.*]] = bitcast i32* [[TMP17]] to i8* +; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, i8* [[TMP18]], i64 3 +; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT]], %struct* [[A]], i64 0, i32 2 +; CHECK-NEXT: store float 0.000000e+00, float* [[TMP20]], align 8 +; CHECK-NEXT: [[TMP21:%.*]] = load i8, i8* [[TMP19]], align 1 +; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT]], %struct* [[A]], i64 0, i32 3 +; CHECK-NEXT: store float 0.000000e+00, float* [[TMP22]], align 4 +; CHECK-NEXT: br label [[BB23]] +; CHECK: bb23: +; CHECK-NEXT: ret void +; +entry: + %tmp1 = fmul float 10.0, 20.0 + %tmp2 = fptosi float %tmp1 to i32 + %tmp3 = fmul float 30.0, 20.0 + %tmp4 = fptosi float %tmp3 to i32 + %tmp5 = icmp sgt i32 100, %tmp2 + %tmp6 = select i1 %tmp5, i32 %tmp2, i32 10 + %tmp7 = select i1 false, i32 0, i32 %tmp6 + %tmp8 = icmp sgt i32 200, %tmp4 + %tmp9 = select i1 %tmp8, i32 %tmp4, i32 300 + %tmp10 = select i1 false, i32 0, i32 %tmp9 + %tmp11 = getelementptr inbounds %struct, %struct* %A, i64 0, i32 0 + store i32 %tmp7, i32* %tmp11, align 8 + %tmp12 = getelementptr inbounds %struct, %struct* %A, i64 0, i32 1 + store i32 %tmp10, i32* %tmp12, align 4 + %tmp13 = load i32*, i32** %B, align 8 + br i1 %c, label %bb23, label %bb14 + +bb14: + %tmp15 = sext i32 %tmp7 to i64 + %tmp16 = add nsw i64 2, %tmp15 + %tmp17 = getelementptr inbounds i32, i32* %tmp13, i64 %tmp16 + %tmp18 = bitcast i32* %tmp17 to i8* + %tmp19 = getelementptr inbounds i8, i8* %tmp18, i64 3 + %tmp20 = getelementptr inbounds %struct, %struct* %A, i64 0, i32 2 + store float 0.0, float* %tmp20, align 8 + %tmp21 = load i8, i8* %tmp19, align 1 + %tmp22 = getelementptr inbounds %struct, %struct* %A, i64 0, i32 3 + store float 0.0, float* %tmp22, align 4 + br label %bb23 + +bb23: + ret void +} + +; In this test there's a single bound, do not generate runtime checks. +define void @single_membound(double* %arg, double* %arg1, double %x) { +; CHECK-LABEL: @single_membound( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP:%.*]] = fsub double [[X:%.*]], 9.900000e+01 +; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds double, double* [[ARG:%.*]], i64 1 +; CHECK-NEXT: store double [[TMP]], double* [[TMP9]], align 8 +; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds double, double* [[ARG1:%.*]], i64 0 +; CHECK-NEXT: [[TMP12:%.*]] = load double, double* [[TMP10]], align 8 +; CHECK-NEXT: [[TMP13:%.*]] = fsub double 1.000000e+00, [[TMP12]] +; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds double, double* [[ARG]], i64 2 +; CHECK-NEXT: br label [[BB15:%.*]] +; CHECK: bb15: +; CHECK-NEXT: [[TMP16:%.*]] = fmul double [[TMP]], 2.000000e+01 +; CHECK-NEXT: store double [[TMP16]], double* [[TMP9]], align 8 +; CHECK-NEXT: [[TMP17:%.*]] = fmul double [[TMP13]], 3.000000e+01 +; CHECK-NEXT: store double [[TMP17]], double* [[TMP14]], align 8 +; CHECK-NEXT: ret void +; +entry: + %tmp = fsub double %x, 99.0 + %tmp9 = getelementptr inbounds double, double* %arg, i64 1 + store double %tmp, double* %tmp9, align 8 + %tmp10 = getelementptr inbounds double, double* %arg1, i64 0 + %tmp12 = load double, double* %tmp10, align 8 + %tmp13 = fsub double 1.0, %tmp12 + %tmp14 = getelementptr inbounds double, double* %arg, i64 2 + br label %bb15 + +bb15: + %tmp16 = fmul double %tmp, 20.0 + store double %tmp16, double* %tmp9, align 8 + %tmp17 = fmul double %tmp13, 30.0 + store double %tmp17, double* %tmp14, align 8 + ret void +} -- 2.7.4