llvm-6502/test/CodeGen/X86/vec_extract-sse4.ll
Evan Cheng 77f0b7a50a Instead of a vector load, shuffle and then extract an element. Load the element from address with an offset.
pshufd $1, (%rdi), %xmm0
        movd %xmm0, %eax
=>
        movl 4(%rdi), %eax


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@51026 91177308-0d34-0410-b5e6-96231b3b80d8
2008-05-13 08:35:03 +00:00

32 lines
784 B
LLVM

; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse41 -o %t -f
; RUN: grep extractps %t | count 1
; RUN: grep pextrd %t | count 1
; RUN: not grep pshufd %t
; RUN: not grep movss %t
define void @t1(float* %R, <4 x float>* %P1) nounwind {
%X = load <4 x float>* %P1
%tmp = extractelement <4 x float> %X, i32 3
store float %tmp, float* %R
ret void
}
define float @t2(<4 x float>* %P1) nounwind {
%X = load <4 x float>* %P1
%tmp = extractelement <4 x float> %X, i32 2
ret float %tmp
}
define void @t3(i32* %R, <4 x i32>* %P1) nounwind {
%X = load <4 x i32>* %P1
%tmp = extractelement <4 x i32> %X, i32 3
store i32 %tmp, i32* %R
ret void
}
define i32 @t4(<4 x i32>* %P1) nounwind {
%X = load <4 x i32>* %P1
%tmp = extractelement <4 x i32> %X, i32 3
ret i32 %tmp
}