2016-02-15 21:41:41 +08:00
|
|
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
|
|
; RUN: llc < %s -mtriple=i386-apple-darwin8 -mattr=+sse2 | FileCheck %s --check-prefix=X32
|
|
|
|
; RUN: llc < %s -mtriple=x86_64-apple-darwin8 -mattr=+sse2 | FileCheck %s --check-prefix=X64
|
when we know the signbit of an input to uint_to_fp is zero,
change it to sint_to_fp on targets where that is cheaper (and
visaversa of course). This allows us to compile uint_to_fp to:
_test:
movl 4(%esp), %eax
shrl $23, %eax
cvtsi2ss %eax, %xmm0
movl 8(%esp), %eax
movss %xmm0, (%eax)
ret
instead of:
.align 3
LCPI1_0: ## double
.long 0 ## double least significant word 4.5036e+15
.long 1127219200 ## double most significant word 4.5036e+15
.text
.align 4,0x90
.globl _test
_test:
subl $12, %esp
movl 16(%esp), %eax
shrl $23, %eax
movl %eax, (%esp)
movl $1127219200, 4(%esp)
movsd (%esp), %xmm0
subsd LCPI1_0, %xmm0
cvtsd2ss %xmm0, %xmm0
movl 20(%esp), %eax
movss %xmm0, (%eax)
addl $12, %esp
ret
llvm-svn: 52747
2008-06-26 08:16:49 +08:00
|
|
|
; rdar://6034396
|
|
|
|
|
2016-02-15 21:41:41 +08:00
|
|
|
define void @test(i32 %x, float* %y) nounwind {
|
|
|
|
; X32-LABEL: test:
|
2017-12-05 01:18:51 +08:00
|
|
|
; X32: ## %bb.0: ## %entry
|
2016-02-15 21:41:41 +08:00
|
|
|
; X32-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
|
|
; X32-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
|
|
; X32-NEXT: shrl $23, %ecx
|
2019-05-07 05:39:51 +08:00
|
|
|
; X32-NEXT: cvtsi2ss %ecx, %xmm0
|
2016-02-15 21:41:41 +08:00
|
|
|
; X32-NEXT: movss %xmm0, (%eax)
|
|
|
|
; X32-NEXT: retl
|
|
|
|
;
|
|
|
|
; X64-LABEL: test:
|
2017-12-05 01:18:51 +08:00
|
|
|
; X64: ## %bb.0: ## %entry
|
2016-02-15 21:41:41 +08:00
|
|
|
; X64-NEXT: shrl $23, %edi
|
2019-05-07 05:39:51 +08:00
|
|
|
; X64-NEXT: cvtsi2ss %edi, %xmm0
|
2016-02-15 21:41:41 +08:00
|
|
|
; X64-NEXT: movss %xmm0, (%rsi)
|
|
|
|
; X64-NEXT: retq
|
when we know the signbit of an input to uint_to_fp is zero,
change it to sint_to_fp on targets where that is cheaper (and
visaversa of course). This allows us to compile uint_to_fp to:
_test:
movl 4(%esp), %eax
shrl $23, %eax
cvtsi2ss %eax, %xmm0
movl 8(%esp), %eax
movss %xmm0, (%eax)
ret
instead of:
.align 3
LCPI1_0: ## double
.long 0 ## double least significant word 4.5036e+15
.long 1127219200 ## double most significant word 4.5036e+15
.text
.align 4,0x90
.globl _test
_test:
subl $12, %esp
movl 16(%esp), %eax
shrl $23, %eax
movl %eax, (%esp)
movl $1127219200, 4(%esp)
movsd (%esp), %xmm0
subsd LCPI1_0, %xmm0
cvtsd2ss %xmm0, %xmm0
movl 20(%esp), %eax
movss %xmm0, (%eax)
addl $12, %esp
ret
llvm-svn: 52747
2008-06-26 08:16:49 +08:00
|
|
|
entry:
|
2016-02-15 21:41:41 +08:00
|
|
|
lshr i32 %x, 23
|
|
|
|
uitofp i32 %0 to float
|
|
|
|
store float %1, float* %y
|
|
|
|
ret void
|
when we know the signbit of an input to uint_to_fp is zero,
change it to sint_to_fp on targets where that is cheaper (and
visaversa of course). This allows us to compile uint_to_fp to:
_test:
movl 4(%esp), %eax
shrl $23, %eax
cvtsi2ss %eax, %xmm0
movl 8(%esp), %eax
movss %xmm0, (%eax)
ret
instead of:
.align 3
LCPI1_0: ## double
.long 0 ## double least significant word 4.5036e+15
.long 1127219200 ## double most significant word 4.5036e+15
.text
.align 4,0x90
.globl _test
_test:
subl $12, %esp
movl 16(%esp), %eax
shrl $23, %eax
movl %eax, (%esp)
movl $1127219200, 4(%esp)
movsd (%esp), %xmm0
subsd LCPI1_0, %xmm0
cvtsd2ss %xmm0, %xmm0
movl 20(%esp), %eax
movss %xmm0, (%eax)
addl $12, %esp
ret
llvm-svn: 52747
2008-06-26 08:16:49 +08:00
|
|
|
}
|