|
1 | 1 | ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3
|
2 |
| -; RUN: opt < %s -passes=slp-vectorizer -S -mtriple=x86_64-unknown-linux-gnu -mcpu=skylake-avx512 | FileCheck %s |
| 2 | +; RUN: opt < %s -passes=slp-vectorizer -S -mtriple=x86_64-unknown-linux-gnu | FileCheck %s |
3 | 3 |
|
4 |
| -define <8 x float> @test(ptr %x, float %v, float %a) { |
5 |
| -; CHECK-LABEL: define <8 x float> @test( |
6 |
| -; CHECK-SAME: ptr [[X:%.*]], float [[V:%.*]], float [[A:%.*]]) #[[ATTR0:[0-9]+]] { |
7 |
| -; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x float> poison, float [[A]], i32 0 |
8 |
| -; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <4 x float> [[TMP2]], <4 x float> poison, <4 x i32> zeroinitializer |
9 |
| -; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x float> poison, float [[V]], i32 0 |
10 |
| -; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <4 x float> [[TMP4]], <4 x float> poison, <4 x i32> zeroinitializer |
11 |
| -; CHECK-NEXT: [[TMP8:%.*]] = fadd <4 x float> [[TMP3]], [[TMP7]] |
12 |
| -; CHECK-NEXT: [[TMP6:%.*]] = load <2 x float>, ptr [[X]], align 4 |
13 |
| -; CHECK-NEXT: [[TMP11:%.*]] = call <4 x float> @llvm.vector.insert.v4f32.v2f32(<4 x float> [[TMP7]], <2 x float> [[TMP6]], i64 0) |
14 |
| -; CHECK-NEXT: [[TMP12:%.*]] = fadd <4 x float> [[TMP3]], [[TMP11]] |
15 |
| -; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <4 x float> [[TMP8]], <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7> |
16 |
| -; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <4 x float> [[TMP12]], <4 x float> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison> |
17 |
| -; CHECK-NEXT: [[I71:%.*]] = shufflevector <8 x float> [[TMP9]], <8 x float> [[TMP10]], <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7> |
18 |
| -; CHECK-NEXT: ret <8 x float> [[I71]] |
| 4 | +define <16 x double> @test(ptr %x, double %v, double %a) { |
| 5 | +; CHECK-LABEL: define <16 x double> @test( |
| 6 | +; CHECK-SAME: ptr [[X:%.*]], double [[V:%.*]], double [[A:%.*]]) { |
| 7 | +; CHECK-NEXT: [[GEP6:%.*]] = getelementptr inbounds double, ptr [[X]], i64 8 |
| 8 | +; CHECK-NEXT: [[GEP8:%.*]] = getelementptr inbounds double, ptr [[X]], i64 9 |
| 9 | +; CHECK-NEXT: [[TMP1:%.*]] = load <6 x double>, ptr [[X]], align 4 |
| 10 | +; CHECK-NEXT: [[TMP6:%.*]] = load <2 x double>, ptr [[GEP6]], align 4 |
| 11 | +; CHECK-NEXT: [[TMP7:%.*]] = load <2 x double>, ptr [[GEP8]], align 4 |
| 12 | +; CHECK-NEXT: [[TMP4:%.*]] = insertelement <16 x double> poison, double [[A]], i32 0 |
| 13 | +; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <16 x double> [[TMP4]], <16 x double> poison, <16 x i32> zeroinitializer |
| 14 | +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x double> poison, double [[V]], i32 0 |
| 15 | +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> poison, <4 x i32> zeroinitializer |
| 16 | +; CHECK-NEXT: [[TMP8:%.*]] = insertelement <2 x double> poison, double [[V]], i32 0 |
| 17 | +; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <2 x double> [[TMP8]], <2 x double> poison, <2 x i32> zeroinitializer |
| 18 | +; CHECK-NEXT: [[TMP10:%.*]] = call <16 x double> @llvm.vector.insert.v16f64.v6f64(<16 x double> poison, <6 x double> [[TMP1]], i64 0) |
| 19 | +; CHECK-NEXT: [[TMP11:%.*]] = shufflevector <4 x double> [[TMP3]], <4 x double> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison> |
| 20 | +; CHECK-NEXT: [[TMP12:%.*]] = shufflevector <16 x double> [[TMP10]], <16 x double> [[TMP11]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 18, i32 19, i32 poison, i32 poison> |
| 21 | +; CHECK-NEXT: [[TMP13:%.*]] = call <16 x double> @llvm.vector.insert.v16f64.v2f64(<16 x double> [[TMP12]], <2 x double> [[TMP6]], i64 6) |
| 22 | +; CHECK-NEXT: [[TMP14:%.*]] = call <16 x double> @llvm.vector.insert.v16f64.v2f64(<16 x double> [[TMP13]], <2 x double> [[TMP7]], i64 8) |
| 23 | +; CHECK-NEXT: [[TMP15:%.*]] = call <16 x double> @llvm.vector.insert.v16f64.v2f64(<16 x double> [[TMP14]], <2 x double> [[TMP9]], i64 10) |
| 24 | +; CHECK-NEXT: [[TMP16:%.*]] = call <16 x double> @llvm.vector.insert.v16f64.v2f64(<16 x double> [[TMP15]], <2 x double> [[TMP9]], i64 12) |
| 25 | +; CHECK-NEXT: [[TMP17:%.*]] = call <16 x double> @llvm.vector.insert.v16f64.v2f64(<16 x double> [[TMP16]], <2 x double> [[TMP9]], i64 14) |
| 26 | +; CHECK-NEXT: [[TMP18:%.*]] = fadd <16 x double> [[TMP5]], [[TMP17]] |
| 27 | +; CHECK-NEXT: ret <16 x double> [[TMP18]] |
19 | 28 | ;
|
20 |
| - %gep1 = getelementptr inbounds <4 x float>, ptr %x, i64 0, i64 1 |
21 |
| - %x0 = load float, ptr %x, align 4 |
22 |
| - %x1 = load float, ptr %gep1, align 4 |
23 |
| - %add1 = fadd float %a, %v |
24 |
| - %add2 = fadd float %a, %v |
25 |
| - %add3 = fadd float %a, %v |
26 |
| - %add4 = fadd float %a, %v |
27 |
| - %add5 = fadd float %a, %x0 |
28 |
| - %add6 = fadd float %a, %x1 |
29 |
| - %add7 = fadd float %a, %v |
30 |
| - %add8 = fadd float %a, %v |
31 |
| - %i0 = insertelement <8 x float> undef, float %add1, i32 0 |
32 |
| - %i1 = insertelement <8 x float> %i0, float %add2, i32 1 |
33 |
| - %i2 = insertelement <8 x float> %i1, float %add3, i32 2 |
34 |
| - %i3 = insertelement <8 x float> %i2, float %add4, i32 3 |
35 |
| - %i4 = insertelement <8 x float> %i3, float %add5, i32 0 |
36 |
| - %i5 = insertelement <8 x float> %i4, float %add6, i32 1 |
37 |
| - %i6 = insertelement <8 x float> %i5, float %add7, i32 2 |
38 |
| - %i7 = insertelement <8 x float> %i6, float %add8, i32 3 |
39 |
| - ret <8 x float> %i7 |
| 29 | + %gep1 = getelementptr inbounds double, ptr %x, i64 1 |
| 30 | + %gep2 = getelementptr inbounds double, ptr %x, i64 2 |
| 31 | + %gep3 = getelementptr inbounds double, ptr %x, i64 3 |
| 32 | + %gep4 = getelementptr inbounds double, ptr %x, i64 4 |
| 33 | + %gep5 = getelementptr inbounds double, ptr %x, i64 5 |
| 34 | + %gep6 = getelementptr inbounds double, ptr %x, i64 8 |
| 35 | + %gep7 = getelementptr inbounds double, ptr %x, i64 9 |
| 36 | + %gep8 = getelementptr inbounds double, ptr %x, i64 9 |
| 37 | + %gep9 = getelementptr inbounds double, ptr %x, i64 10 |
| 38 | + %x0 = load double, ptr %x, align 4 |
| 39 | + %x1 = load double, ptr %gep1, align 4 |
| 40 | + %x2 = load double, ptr %gep2, align 4 |
| 41 | + %x3 = load double, ptr %gep3, align 4 |
| 42 | + %x4 = load double, ptr %gep4, align 4 |
| 43 | + %x5 = load double, ptr %gep5, align 4 |
| 44 | + %x6 = load double, ptr %gep6, align 4 |
| 45 | + %x7 = load double, ptr %gep7, align 4 |
| 46 | + %x8 = load double, ptr %gep8, align 4 |
| 47 | + %x9 = load double, ptr %gep9, align 4 |
| 48 | + %add1 = fadd double %a, %x0 |
| 49 | + %add2 = fadd double %a, %x1 |
| 50 | + %add3 = fadd double %a, %x2 |
| 51 | + %add4 = fadd double %a, %x3 |
| 52 | + %add5 = fadd double %a, %x4 |
| 53 | + %add6 = fadd double %a, %x5 |
| 54 | + %add7 = fadd double %a, %x6 |
| 55 | + %add8 = fadd double %a, %x7 |
| 56 | + %add9 = fadd double %a, %x8 |
| 57 | + %add10 = fadd double %a, %x9 |
| 58 | + %add11 = fadd double %a, %v |
| 59 | + %add12 = fadd double %a, %v |
| 60 | + %add13 = fadd double %a, %v |
| 61 | + %add14 = fadd double %a, %v |
| 62 | + %add15 = fadd double %a, %v |
| 63 | + %add16 = fadd double %a, %v |
| 64 | + %i0 = insertelement <16 x double> poison, double %add1, i32 0 |
| 65 | + %i1 = insertelement <16 x double> %i0, double %add2, i32 1 |
| 66 | + %i2 = insertelement <16 x double> %i1, double %add3, i32 2 |
| 67 | + %i3 = insertelement <16 x double> %i2, double %add4, i32 3 |
| 68 | + %i4 = insertelement <16 x double> %i3, double %add5, i32 4 |
| 69 | + %i5 = insertelement <16 x double> %i4, double %add6, i32 5 |
| 70 | + %i6 = insertelement <16 x double> %i5, double %add7, i32 6 |
| 71 | + %i7 = insertelement <16 x double> %i6, double %add8, i32 7 |
| 72 | + %i8 = insertelement <16 x double> %i7, double %add9, i32 8 |
| 73 | + %i9 = insertelement <16 x double> %i8, double %add10, i32 9 |
| 74 | + %i10 = insertelement <16 x double> %i9, double %add11, i32 10 |
| 75 | + %i11 = insertelement <16 x double> %i10, double %add12, i32 11 |
| 76 | + %i12 = insertelement <16 x double> %i11, double %add13, i32 12 |
| 77 | + %i13 = insertelement <16 x double> %i12, double %add14, i32 13 |
| 78 | + %i14 = insertelement <16 x double> %i13, double %add15, i32 14 |
| 79 | + %i15 = insertelement <16 x double> %i14, double %add16, i32 15 |
| 80 | + ret <16 x double> %i15 |
40 | 81 | }
|
0 commit comments