mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-21 03:53:04 +02:00
99f7e3a3dd
This associates movss and movsd with the packed single and packed double execution domains (resp.). While this is largely cosmetic, as we now don't have weird ping-pong-ing between single and double precision, it is also useful because it avoids the domain fixing algorithm from seeing domain breaks that don't actually exist. It will also be much more important if we have an execution domain default other than packed single, as that would cause us to mix movss and movsd with integer vector code on a regular basis, a very bad mixture. llvm-svn: 228135
44 lines
1.2 KiB
LLVM
44 lines
1.2 KiB
LLVM
; RUN: llc < %s -mtriple=i386-unknown-unknown -march=x86 -mattr=+sse2 | FileCheck %s
|
|
|
|
; rdar://6504833
|
|
define float @test1(i32 %x) nounwind readnone {
|
|
; CHECK-LABEL: test1:
|
|
; CHECK: # BB#0: # %entry
|
|
; CHECK-NEXT: pushl %eax
|
|
; CHECK-NEXT: movsd .LCPI0_0, %xmm0
|
|
; CHECK-NEXT: movd {{[0-9]+}}(%esp), %xmm1
|
|
; CHECK-NEXT: orpd %xmm0, %xmm1
|
|
; CHECK-NEXT: subsd %xmm0, %xmm1
|
|
; CHECK-NEXT: xorps %xmm0, %xmm0
|
|
; CHECK-NEXT: cvtsd2ss %xmm1, %xmm0
|
|
; CHECK-NEXT: movss %xmm0, (%esp)
|
|
; CHECK-NEXT: flds (%esp)
|
|
; CHECK-NEXT: popl %eax
|
|
; CHECK-NEXT: retl
|
|
entry:
|
|
%0 = uitofp i32 %x to float
|
|
ret float %0
|
|
}
|
|
|
|
; PR10802
|
|
define float @test2(<4 x i32> %x) nounwind readnone ssp {
|
|
; CHECK-LABEL: test2:
|
|
; CHECK: # BB#0: # %entry
|
|
; CHECK-NEXT: pushl %eax
|
|
; CHECK-NEXT: xorps %xmm1, %xmm1
|
|
; CHECK-NEXT: movss %xmm0, %xmm1
|
|
; CHECK-NEXT: movsd .LCPI1_0, %xmm0
|
|
; CHECK-NEXT: orps %xmm0, %xmm1
|
|
; CHECK-NEXT: subsd %xmm0, %xmm1
|
|
; CHECK-NEXT: xorps %xmm0, %xmm0
|
|
; CHECK-NEXT: cvtsd2ss %xmm1, %xmm0
|
|
; CHECK-NEXT: movss %xmm0, (%esp)
|
|
; CHECK-NEXT: flds (%esp)
|
|
; CHECK-NEXT: popl %eax
|
|
; CHECK-NEXT: retl
|
|
entry:
|
|
%vecext = extractelement <4 x i32> %x, i32 0
|
|
%conv = uitofp i32 %vecext to float
|
|
ret float %conv
|
|
}
|