8d75f97662
Need to check the first split node as a potential reduction root to prevent compiler crash
41 lines
1.9 KiB
LLVM
41 lines
1.9 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: opt -S --passes=slp-vectorizer -mtriple=x86_64-unknown-linux-gnu < %s | FileCheck %s
|
|
|
|
define i64 @test(i64 %0) {
|
|
; CHECK-LABEL: define i64 @test(
|
|
; CHECK-SAME: i64 [[TMP0:%.*]]) {
|
|
; CHECK-NEXT: [[BB:.*:]]
|
|
; CHECK-NEXT: [[TMP1:%.*]] = alloca [20 x i64], align 8
|
|
; CHECK-NEXT: store i64 0, ptr [[TMP1]], align 8
|
|
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP1]], i64 72
|
|
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <2 x i64> poison, i64 [[TMP0]], i32 0
|
|
; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <2 x i64> [[TMP3]], <2 x i64> poison, <2 x i32> zeroinitializer
|
|
; CHECK-NEXT: [[TMP5:%.*]] = or <2 x i64> [[TMP4]], splat (i64 1)
|
|
; CHECK-NEXT: [[TMP6:%.*]] = load <4 x i64>, ptr [[TMP2]], align 8
|
|
; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <4 x i64> [[TMP6]], <4 x i64> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison>
|
|
; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <2 x i64> [[TMP5]], <2 x i64> poison, <6 x i32> <i32 0, i32 1, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <6 x i64> [[TMP7]], <6 x i64> [[TMP8]], <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 6, i32 7>
|
|
; CHECK-NEXT: [[TMP10:%.*]] = call i64 @llvm.vector.reduce.add.v6i64(<6 x i64> [[TMP9]])
|
|
; CHECK-NEXT: ret i64 [[TMP10]]
|
|
;
|
|
bb:
|
|
%1 = alloca [20 x i64], align 8
|
|
store i64 0, ptr %1, align 8
|
|
%2 = getelementptr i8, ptr %1, i64 88
|
|
%3 = getelementptr i8, ptr %1, i64 80
|
|
%4 = load i64, ptr %3, align 8
|
|
%5 = getelementptr i8, ptr %1, i64 72
|
|
%6 = load i64, ptr %5, align 8
|
|
%7 = or i64 %0, 1
|
|
%8 = or i64 %0, 1
|
|
%9 = add i64 %7, %8
|
|
%10 = add i64 %6, %9
|
|
%11 = getelementptr i8, ptr %1, i64 96
|
|
%12 = load i64, ptr %11, align 8
|
|
%13 = load i64, ptr %2, align 8
|
|
%14 = add i64 %4, %10
|
|
%15 = add i64 %13, %14
|
|
%16 = add i64 %12, %15
|
|
ret i64 %16
|
|
}
|