|
| 1 | +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py |
| 2 | +; RUN: opt -passes=slp-vectorizer -S < %s -mtriple=x86_64-unknown-linux -slp-threshold=-160 | FileCheck %s |
| 3 | + |
| 4 | +define void @test1(i128 %p0, i128 %p1, i128 %p2, i128 %p3, <4 x i128> %vec) { |
| 5 | +; CHECK-LABEL: @test1( |
| 6 | +; CHECK-NEXT: entry: |
| 7 | +; CHECK-NEXT: [[TMP0:%.*]] = insertelement <2 x i128> poison, i128 [[P0:%.*]], i32 0 |
| 8 | +; CHECK-NEXT: [[TMP1:%.*]] = insertelement <2 x i128> [[TMP0]], i128 [[P1:%.*]], i32 1 |
| 9 | +; CHECK-NEXT: [[TMP2:%.*]] = trunc <2 x i128> [[TMP1]] to <2 x i32> |
| 10 | +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <2 x i32> [[TMP2]], <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 0, i32 1> |
| 11 | +; CHECK-NEXT: [[TMP4:%.*]] = insertelement <2 x i128> poison, i128 [[P2:%.*]], i32 0 |
| 12 | +; CHECK-NEXT: [[TMP5:%.*]] = insertelement <2 x i128> [[TMP4]], i128 [[P3:%.*]], i32 1 |
| 13 | +; CHECK-NEXT: [[TMP6:%.*]] = trunc <2 x i128> [[TMP5]] to <2 x i32> |
| 14 | +; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <2 x i32> [[TMP6]], <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 0, i32 1> |
| 15 | +; CHECK-NEXT: [[T5:%.*]] = trunc i128 [[P1]] to i32 |
| 16 | +; CHECK-NEXT: [[TMP8:%.*]] = sdiv <4 x i32> [[TMP3]], [[TMP7]] |
| 17 | +; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <2 x i128> [[TMP1]], <2 x i128> [[TMP5]], <4 x i32> <i32 poison, i32 0, i32 3, i32 2> |
| 18 | +; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <4 x i128> [[VEC:%.*]], <4 x i128> [[TMP9]], <4 x i32> <i32 0, i32 5, i32 6, i32 7> |
| 19 | +; CHECK-NEXT: [[TMP11:%.*]] = trunc <4 x i128> [[TMP10]] to <4 x i32> |
| 20 | +; CHECK-NEXT: [[TMP12:%.*]] = sdiv <4 x i32> [[TMP8]], [[TMP11]] |
| 21 | +; CHECK-NEXT: br label [[BB:%.*]] |
| 22 | +; CHECK: bb: |
| 23 | +; CHECK-NEXT: [[TMP13:%.*]] = phi <4 x i32> [ [[TMP12]], [[ENTRY:%.*]] ] |
| 24 | +; CHECK-NEXT: ret void |
| 25 | +; |
| 26 | +entry: |
| 27 | + %t1 = trunc i128 %p0 to i32 |
| 28 | + %t2 = trunc i128 %p1 to i32 |
| 29 | + %t3 = trunc i128 %p2 to i32 |
| 30 | + %t4 = trunc i128 %p3 to i32 |
| 31 | + %t5 = trunc i128 %p1 to i32 |
| 32 | + %t6 = trunc i128 %p0 to i32 |
| 33 | + %t7 = trunc i128 %p3 to i32 |
| 34 | + %t8 = trunc i128 %p2 to i32 |
| 35 | + %m0 = sdiv i32 %t1, %t3 |
| 36 | + %m1 = sdiv i32 %t2, %t4 |
| 37 | + %m2 = sdiv i32 %t1, %t3 |
| 38 | + %m3 = sdiv i32 %t2, %t4 |
| 39 | + %e0 = extractelement <4 x i128> %vec, i32 0 |
| 40 | + %t9 = trunc i128 %e0 to i32 |
| 41 | + %d0 = sdiv i32 %m0, %t9 |
| 42 | + %d1 = sdiv i32 %m1, %t6 |
| 43 | + %d2 = sdiv i32 %m2, %t7 |
| 44 | + %d3 = sdiv i32 %m3, %t8 |
| 45 | + br label %bb |
| 46 | + |
| 47 | +bb: |
| 48 | + %phi0 = phi i32 [ %d0, %entry ] |
| 49 | + %phi1 = phi i32 [ %d1, %entry ] |
| 50 | + %phi2 = phi i32 [ %d2, %entry ] |
| 51 | + %phi3 = phi i32 [ %d3, %entry ] |
| 52 | + ret void |
| 53 | +} |
0 commit comments