1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 04:02:41 +01:00
llvm-mirror/test/CodeGen/X86/vec_cast3.ll
Craig Topper 39e018707b [X86] In FP_TO_INTHelper, when moving data from SSE register to X87 register file via the stack, use the same stack slot we use for the integer conversion.
No need for a separate stack slot. The lifetimes don't overlap.

Also fix the MachinePointerInfo for the final load after the integer conversion to indicate it came from the stack slot.

llvm-svn: 354234
2019-02-17 19:23:49 +00:00

264 lines
10 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i386-apple-darwin10 -mattr=+avx | FileCheck %s
; RUN: llc < %s -mtriple=i386-apple-darwin10 -mattr=+avx -x86-experimental-vector-widening-legalization | FileCheck %s --check-prefix=CHECK-WIDE
define <2 x float> @cvt_v2i8_v2f32(<2 x i8> %src) {
; CHECK-LABEL: cvt_v2i8_v2f32:
; CHECK: ## %bb.0:
; CHECK-NEXT: vpsllq $56, %xmm0, %xmm0
; CHECK-NEXT: vpsrad $24, %xmm0, %xmm0
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
; CHECK-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2i8_v2f32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vpmovsxbd %xmm0, %xmm0
; CHECK-WIDE-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-WIDE-NEXT: retl
%res = sitofp <2 x i8> %src to <2 x float>
ret <2 x float> %res
}
define <2 x float> @cvt_v2i16_v2f32(<2 x i16> %src) {
; CHECK-LABEL: cvt_v2i16_v2f32:
; CHECK: ## %bb.0:
; CHECK-NEXT: vpsllq $48, %xmm0, %xmm0
; CHECK-NEXT: vpsrad $16, %xmm0, %xmm0
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
; CHECK-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2i16_v2f32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vpmovsxwd %xmm0, %xmm0
; CHECK-WIDE-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-WIDE-NEXT: retl
%res = sitofp <2 x i16> %src to <2 x float>
ret <2 x float> %res
}
define <2 x float> @cvt_v2i32_v2f32(<2 x i32> %src) {
; CHECK-LABEL: cvt_v2i32_v2f32:
; CHECK: ## %bb.0:
; CHECK-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
; CHECK-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2i32_v2f32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-WIDE-NEXT: retl
%res = sitofp <2 x i32> %src to <2 x float>
ret <2 x float> %res
}
define <2 x float> @cvt_v2u8_v2f32(<2 x i8> %src) {
; CHECK-LABEL: cvt_v2u8_v2f32:
; CHECK: ## %bb.0:
; CHECK-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[u,u,u,u,u,u,u,u]
; CHECK-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2u8_v2f32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
; CHECK-WIDE-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-WIDE-NEXT: retl
%res = uitofp <2 x i8> %src to <2 x float>
ret <2 x float> %res
}
define <2 x float> @cvt_v2u16_v2f32(<2 x i16> %src) {
; CHECK-LABEL: cvt_v2u16_v2f32:
; CHECK: ## %bb.0:
; CHECK-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1],zero,zero,xmm0[8,9],zero,zero,xmm0[8,9],zero,zero,xmm0[10,11],zero,zero
; CHECK-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2u16_v2f32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
; CHECK-WIDE-NEXT: vcvtdq2ps %xmm0, %xmm0
; CHECK-WIDE-NEXT: retl
%res = uitofp <2 x i16> %src to <2 x float>
ret <2 x float> %res
}
define <2 x float> @cvt_v2u32_v2f32(<2 x i32> %src) {
; CHECK-LABEL: cvt_v2u32_v2f32:
; CHECK: ## %bb.0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; CHECK-NEXT: vmovaps {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; CHECK-NEXT: vorps %xmm1, %xmm0, %xmm0
; CHECK-NEXT: vsubpd %xmm1, %xmm0, %xmm0
; CHECK-NEXT: vcvtpd2ps %xmm0, %xmm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2u32_v2f32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
; CHECK-WIDE-NEXT: vmovdqa {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; CHECK-WIDE-NEXT: vpor %xmm1, %xmm0, %xmm0
; CHECK-WIDE-NEXT: vsubpd %xmm1, %xmm0, %xmm0
; CHECK-WIDE-NEXT: vcvtpd2ps %xmm0, %xmm0
; CHECK-WIDE-NEXT: retl
%res = uitofp <2 x i32> %src to <2 x float>
ret <2 x float> %res
}
define <2 x i8> @cvt_v2f32_v2i8(<2 x float> %src) {
; CHECK-LABEL: cvt_v2f32_v2i8:
; CHECK: ## %bb.0:
; CHECK-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-NEXT: vpmovsxdq %xmm0, %xmm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2f32_v2i8:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-WIDE-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; CHECK-WIDE-NEXT: retl
%res = fptosi <2 x float> %src to <2 x i8>
ret <2 x i8> %res
}
define <2 x i16> @cvt_v2f32_v2i16(<2 x float> %src) {
; CHECK-LABEL: cvt_v2f32_v2i16:
; CHECK: ## %bb.0:
; CHECK-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-NEXT: vpmovsxdq %xmm0, %xmm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2f32_v2i16:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-WIDE-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
; CHECK-WIDE-NEXT: retl
%res = fptosi <2 x float> %src to <2 x i16>
ret <2 x i16> %res
}
define <2 x i32> @cvt_v2f32_v2i32(<2 x float> %src) {
; CHECK-LABEL: cvt_v2f32_v2i32:
; CHECK: ## %bb.0:
; CHECK-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2f32_v2i32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-WIDE-NEXT: retl
%res = fptosi <2 x float> %src to <2 x i32>
ret <2 x i32> %res
}
define <2 x i8> @cvt_v2f32_v2u8(<2 x float> %src) {
; CHECK-LABEL: cvt_v2f32_v2u8:
; CHECK: ## %bb.0:
; CHECK-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2f32_v2u8:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-WIDE-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; CHECK-WIDE-NEXT: retl
%res = fptoui <2 x float> %src to <2 x i8>
ret <2 x i8> %res
}
define <2 x i16> @cvt_v2f32_v2u16(<2 x float> %src) {
; CHECK-LABEL: cvt_v2f32_v2u16:
; CHECK: ## %bb.0:
; CHECK-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2f32_v2u16:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-WIDE-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
; CHECK-WIDE-NEXT: retl
%res = fptoui <2 x float> %src to <2 x i16>
ret <2 x i16> %res
}
define <2 x i32> @cvt_v2f32_v2u32(<2 x float> %src) {
; CHECK-LABEL: cvt_v2f32_v2u32:
; CHECK: ## %bb.0:
; CHECK-NEXT: subl $36, %esp
; CHECK-NEXT: .cfi_def_cfa_offset 40
; CHECK-NEXT: vmovshdup {{.*#+}} xmm2 = xmm0[1,1,3,3]
; CHECK-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; CHECK-NEXT: vucomiss %xmm1, %xmm2
; CHECK-NEXT: jb LBB11_2
; CHECK-NEXT: ## %bb.1:
; CHECK-NEXT: vsubss %xmm1, %xmm2, %xmm2
; CHECK-NEXT: LBB11_2:
; CHECK-NEXT: vmovss %xmm2, (%esp)
; CHECK-NEXT: flds (%esp)
; CHECK-NEXT: fisttpll (%esp)
; CHECK-NEXT: setae %al
; CHECK-NEXT: movzbl %al, %eax
; CHECK-NEXT: shll $31, %eax
; CHECK-NEXT: xorl {{[0-9]+}}(%esp), %eax
; CHECK-NEXT: vucomiss %xmm1, %xmm0
; CHECK-NEXT: jb LBB11_4
; CHECK-NEXT: ## %bb.3:
; CHECK-NEXT: vsubss %xmm1, %xmm0, %xmm0
; CHECK-NEXT: LBB11_4:
; CHECK-NEXT: vmovss %xmm0, {{[0-9]+}}(%esp)
; CHECK-NEXT: flds {{[0-9]+}}(%esp)
; CHECK-NEXT: fisttpll {{[0-9]+}}(%esp)
; CHECK-NEXT: setae %cl
; CHECK-NEXT: movzbl %cl, %ecx
; CHECK-NEXT: shll $31, %ecx
; CHECK-NEXT: xorl {{[0-9]+}}(%esp), %ecx
; CHECK-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; CHECK-NEXT: vpinsrd $1, %ecx, %xmm0, %xmm0
; CHECK-NEXT: vpinsrd $2, (%esp), %xmm0, %xmm0
; CHECK-NEXT: vpinsrd $3, %eax, %xmm0, %xmm0
; CHECK-NEXT: addl $36, %esp
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: cvt_v2f32_v2u32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vmovaps {{.*#+}} xmm1 = [2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9]
; CHECK-WIDE-NEXT: vcmpltps %xmm1, %xmm0, %xmm2
; CHECK-WIDE-NEXT: vsubps %xmm1, %xmm0, %xmm1
; CHECK-WIDE-NEXT: vcvttps2dq %xmm1, %xmm1
; CHECK-WIDE-NEXT: vxorps LCPI11_1, %xmm1, %xmm1
; CHECK-WIDE-NEXT: vcvttps2dq %xmm0, %xmm0
; CHECK-WIDE-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; CHECK-WIDE-NEXT: retl
%res = fptoui <2 x float> %src to <2 x i32>
ret <2 x i32> %res
}
define <32 x i8> @PR40146(<4 x i64> %x) {
; CHECK-LABEL: PR40146:
; CHECK: ## %bb.0:
; CHECK-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; CHECK-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; CHECK-NEXT: vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
; CHECK-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; CHECK-NEXT: retl
;
; CHECK-WIDE-LABEL: PR40146:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; CHECK-WIDE-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; CHECK-WIDE-NEXT: vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
; CHECK-WIDE-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; CHECK-WIDE-NEXT: retl
%perm = shufflevector <4 x i64> %x, <4 x i64> undef, <4 x i32> <i32 0, i32 undef, i32 1, i32 undef>
%t1 = bitcast <4 x i64> %perm to <32 x i8>
%t2 = shufflevector <32 x i8> %t1, <32 x i8> <i8 0, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 0, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <32 x i32> <i32 0, i32 32, i32 1, i32 32, i32 2, i32 32, i32 3, i32 32, i32 4, i32 32, i32 5, i32 32, i32 6, i32 32, i32 7, i32 32, i32 16, i32 48, i32 17, i32 48, i32 18, i32 48, i32 19, i32 48, i32 20, i32 48, i32 21, i32 48, i32 22, i32 48, i32 23, i32 48>
ret <32 x i8> %t2
}