forked from OSchip/llvm-project
27 lines
1.1 KiB
LLVM
27 lines
1.1 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mcpu=corei7 -mtriple=x86_64-unknown-linux-gnu | FileCheck %s
|
|
|
|
define void @foo(<4 x float> %in, <4 x i8>* %out) {
|
|
; CHECK-LABEL: foo:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: cvttps2dq %xmm0, %xmm0
|
|
; CHECK-NEXT: pextrb $8, %xmm0, %eax
|
|
; CHECK-NEXT: pextrb $4, %xmm0, %ecx
|
|
; CHECK-NEXT: pextrb $0, %xmm0, %edx
|
|
; CHECK-NEXT: movd %edx, %xmm0
|
|
; CHECK-NEXT: pinsrb $1, %ecx, %xmm0
|
|
; CHECK-NEXT: pinsrb $2, %eax, %xmm0
|
|
; CHECK-NEXT: movl $255, %eax
|
|
; CHECK-NEXT: pinsrb $3, %eax, %xmm0
|
|
; CHECK-NEXT: movd %xmm0, (%rdi)
|
|
; CHECK-NEXT: retq
|
|
%t0 = fptosi <4 x float> %in to <4 x i32>
|
|
%t1 = trunc <4 x i32> %t0 to <4 x i16>
|
|
%t2 = shufflevector <4 x i16> %t1, <4 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
|
|
%t3 = trunc <8 x i16> %t2 to <8 x i8>
|
|
%t4 = shufflevector <8 x i8> %t3, <8 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
|
|
%t5 = insertelement <4 x i8> %t4, i8 -1, i32 3
|
|
store <4 x i8> %t5, <4 x i8>* %out
|
|
ret void
|
|
}
|