07507cb591
Need to choose the size of vector factor for mask based on the entries vector factors, not mask size, to generate correct code. Fixes #117170
56 lines
2.6 KiB
LLVM
56 lines
2.6 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: %if x86-registered-target %{ opt -passes=slp-vectorizer -S < %s -mtriple=x86_64-unknown-linux -slp-threshold=-165 | FileCheck %s %}
|
|
; RUN: %if aarch64-registered-target %{ opt -passes=slp-vectorizer -S < %s -mtriple=aarch64-unknown-linux -slp-threshold=-165 | FileCheck %s %}
|
|
|
|
define void @test1(i128 %p0, i128 %p1, i128 %p2, i128 %p3, <4 x i128> %vec) {
|
|
; CHECK-LABEL: @test1(
|
|
; CHECK-NEXT: entry:
|
|
; CHECK-NEXT: [[TMP0:%.*]] = insertelement <2 x i128> poison, i128 [[P0:%.*]], i32 0
|
|
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <2 x i128> [[TMP0]], i128 [[P1:%.*]], i32 1
|
|
; CHECK-NEXT: [[TMP2:%.*]] = trunc <2 x i128> [[TMP1]] to <2 x i32>
|
|
; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <2 x i32> [[TMP2]], <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
|
|
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <2 x i128> poison, i128 [[P2:%.*]], i32 0
|
|
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <2 x i128> [[TMP4]], i128 [[P3:%.*]], i32 1
|
|
; CHECK-NEXT: [[TMP6:%.*]] = trunc <2 x i128> [[TMP5]] to <2 x i32>
|
|
; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <2 x i32> [[TMP6]], <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
|
|
; CHECK-NEXT: [[T5:%.*]] = trunc i128 [[P1]] to i32
|
|
; CHECK-NEXT: [[TMP8:%.*]] = sdiv <4 x i32> [[TMP3]], [[TMP7]]
|
|
; CHECK-NEXT: [[TMP9:%.*]] = trunc <4 x i128> [[VEC:%.*]] to <4 x i32>
|
|
; CHECK-NEXT: [[TMP10:%.*]] = trunc <4 x i128> [[VEC]] to <4 x i32>
|
|
; CHECK-NEXT: [[TMP11:%.*]] = shufflevector <2 x i32> [[TMP2]], <2 x i32> [[TMP6]], <4 x i32> <i32 poison, i32 0, i32 3, i32 2>
|
|
; CHECK-NEXT: [[TMP12:%.*]] = shufflevector <4 x i32> [[TMP10]], <4 x i32> [[TMP11]], <4 x i32> <i32 0, i32 5, i32 6, i32 7>
|
|
; CHECK-NEXT: [[TMP13:%.*]] = sdiv <4 x i32> [[TMP8]], [[TMP12]]
|
|
; CHECK-NEXT: br label [[BB:%.*]]
|
|
; CHECK: bb:
|
|
; CHECK-NEXT: [[TMP14:%.*]] = phi <4 x i32> [ [[TMP13]], [[ENTRY:%.*]] ]
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
%t1 = trunc i128 %p0 to i32
|
|
%t2 = trunc i128 %p1 to i32
|
|
%t3 = trunc i128 %p2 to i32
|
|
%t4 = trunc i128 %p3 to i32
|
|
%t5 = trunc i128 %p1 to i32
|
|
%t6 = trunc i128 %p0 to i32
|
|
%t7 = trunc i128 %p3 to i32
|
|
%t8 = trunc i128 %p2 to i32
|
|
%m0 = sdiv i32 %t1, %t3
|
|
%m1 = sdiv i32 %t2, %t4
|
|
%m2 = sdiv i32 %t1, %t3
|
|
%m3 = sdiv i32 %t2, %t4
|
|
%e0 = extractelement <4 x i128> %vec, i32 0
|
|
%t9 = trunc i128 %e0 to i32
|
|
%d0 = sdiv i32 %m0, %t9
|
|
%d1 = sdiv i32 %m1, %t6
|
|
%d2 = sdiv i32 %m2, %t7
|
|
%d3 = sdiv i32 %m3, %t8
|
|
br label %bb
|
|
|
|
bb:
|
|
%phi0 = phi i32 [ %d0, %entry ]
|
|
%phi1 = phi i32 [ %d1, %entry ]
|
|
%phi2 = phi i32 [ %d2, %entry ]
|
|
%phi3 = phi i32 [ %d3, %entry ]
|
|
ret void
|
|
}
|