• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: llc < %s -mtriple=i686-apple-darwin -mattr=+sse2 | FileCheck %s
3
4define <2 x i64> @test_x86_sse2_psll_dq_bs(<2 x i64> %a0) {
5; CHECK-LABEL: test_x86_sse2_psll_dq_bs:
6; CHECK:       ## BB#0:
7; CHECK-NEXT:    pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7,8]
8; CHECK-NEXT:    retl
9  %res = call <2 x i64> @llvm.x86.sse2.psll.dq.bs(<2 x i64> %a0, i32 7) ; <<2 x i64>> [#uses=1]
10  ret <2 x i64> %res
11}
12declare <2 x i64> @llvm.x86.sse2.psll.dq.bs(<2 x i64>, i32) nounwind readnone
13
14
15define <2 x i64> @test_x86_sse2_psrl_dq_bs(<2 x i64> %a0) {
16; CHECK-LABEL: test_x86_sse2_psrl_dq_bs:
17; CHECK:       ## BB#0:
18; CHECK-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero
19; CHECK-NEXT:    retl
20  %res = call <2 x i64> @llvm.x86.sse2.psrl.dq.bs(<2 x i64> %a0, i32 7) ; <<2 x i64>> [#uses=1]
21  ret <2 x i64> %res
22}
23declare <2 x i64> @llvm.x86.sse2.psrl.dq.bs(<2 x i64>, i32) nounwind readnone
24
25define <2 x i64> @test_x86_sse2_psll_dq(<2 x i64> %a0) {
26; CHECK-LABEL: test_x86_sse2_psll_dq:
27; CHECK:       ## BB#0:
28; CHECK-NEXT:    pslldq {{.*#+}} xmm0 = zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14]
29; CHECK-NEXT:    retl
30  %res = call <2 x i64> @llvm.x86.sse2.psll.dq(<2 x i64> %a0, i32 8) ; <<2 x i64>> [#uses=1]
31  ret <2 x i64> %res
32}
33declare <2 x i64> @llvm.x86.sse2.psll.dq(<2 x i64>, i32) nounwind readnone
34
35
36define <2 x i64> @test_x86_sse2_psrl_dq(<2 x i64> %a0) {
37; CHECK-LABEL: test_x86_sse2_psrl_dq:
38; CHECK:       ## BB#0:
39; CHECK-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15],zero
40; CHECK-NEXT:    retl
41  %res = call <2 x i64> @llvm.x86.sse2.psrl.dq(<2 x i64> %a0, i32 8) ; <<2 x i64>> [#uses=1]
42  ret <2 x i64> %res
43}
44declare <2 x i64> @llvm.x86.sse2.psrl.dq(<2 x i64>, i32) nounwind readnone
45
46
47define <2 x double> @test_x86_sse2_cvtdq2pd(<4 x i32> %a0) {
48; CHECK-LABEL: test_x86_sse2_cvtdq2pd:
49; CHECK:       ## BB#0:
50; CHECK-NEXT:    cvtdq2pd %xmm0, %xmm0
51; CHECK-NEXT:    retl
52  %res = call <2 x double> @llvm.x86.sse2.cvtdq2pd(<4 x i32> %a0) ; <<2 x double>> [#uses=1]
53  ret <2 x double> %res
54}
55declare <2 x double> @llvm.x86.sse2.cvtdq2pd(<4 x i32>) nounwind readnone
56
57
58define <2 x double> @test_x86_sse2_cvtps2pd(<4 x float> %a0) {
59; CHECK-LABEL: test_x86_sse2_cvtps2pd:
60; CHECK:       ## BB#0:
61; CHECK-NEXT:    cvtps2pd %xmm0, %xmm0
62; CHECK-NEXT:    retl
63  %res = call <2 x double> @llvm.x86.sse2.cvtps2pd(<4 x float> %a0) ; <<2 x double>> [#uses=1]
64  ret <2 x double> %res
65}
66declare <2 x double> @llvm.x86.sse2.cvtps2pd(<4 x float>) nounwind readnone
67
68
69define <4 x i32> @test_x86_sse2_cvttps2dq(<4 x float> %a0) {
70; CHECK-LABEL: test_x86_sse2_cvttps2dq:
71; CHECK:       ## BB#0:
72; CHECK-NEXT:    cvttps2dq %xmm0, %xmm0
73; CHECK-NEXT:    retl
74  %res = call <4 x i32> @llvm.x86.sse2.cvttps2dq(<4 x float> %a0) ; <<4 x i32>> [#uses=1]
75  ret <4 x i32> %res
76}
77declare <4 x i32> @llvm.x86.sse2.cvttps2dq(<4 x float>) nounwind readnone
78
79
80define void @test_x86_sse2_storel_dq(i8* %a0, <4 x i32> %a1) {
81; CHECK-LABEL: test_x86_sse2_storel_dq:
82; CHECK:       ## BB#0:
83; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
84; CHECK-NEXT:    movlps %xmm0, (%eax)
85; CHECK-NEXT:    retl
86  call void @llvm.x86.sse2.storel.dq(i8* %a0, <4 x i32> %a1)
87  ret void
88}
89declare void @llvm.x86.sse2.storel.dq(i8*, <4 x i32>) nounwind
90
91
92define void @test_x86_sse2_storeu_dq(i8* %a0, <16 x i8> %a1) {
93  ; add operation forces the execution domain.
94; CHECK-LABEL: test_x86_sse2_storeu_dq:
95; CHECK:       ## BB#0:
96; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
97; CHECK-NEXT:    paddb LCPI8_0, %xmm0
98; CHECK-NEXT:    movdqu %xmm0, (%eax)
99; CHECK-NEXT:    retl
100  %a2 = add <16 x i8> %a1, <i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1>
101  call void @llvm.x86.sse2.storeu.dq(i8* %a0, <16 x i8> %a2)
102  ret void
103}
104declare void @llvm.x86.sse2.storeu.dq(i8*, <16 x i8>) nounwind
105
106
107define void @test_x86_sse2_storeu_pd(i8* %a0, <2 x double> %a1) {
108  ; fadd operation forces the execution domain.
109; CHECK-LABEL: test_x86_sse2_storeu_pd:
110; CHECK:       ## BB#0:
111; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
112; CHECK-NEXT:    movsd {{.*#+}} xmm1 = mem[0],zero
113; CHECK-NEXT:    pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1,2,3,4,5,6,7]
114; CHECK-NEXT:    addpd %xmm0, %xmm1
115; CHECK-NEXT:    movupd %xmm1, (%eax)
116; CHECK-NEXT:    retl
117  %a2 = fadd <2 x double> %a1, <double 0x0, double 0x4200000000000000>
118  call void @llvm.x86.sse2.storeu.pd(i8* %a0, <2 x double> %a2)
119  ret void
120}
121declare void @llvm.x86.sse2.storeu.pd(i8*, <2 x double>) nounwind
122
123define <4 x i32> @test_x86_sse2_pshuf_d(<4 x i32> %a) {
124; CHECK-LABEL: test_x86_sse2_pshuf_d:
125; CHECK:       ## BB#0: ## %entry
126; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,2,1,0]
127; CHECK-NEXT:    retl
128entry:
129  %res = call <4 x i32> @llvm.x86.sse2.pshuf.d(<4 x i32> %a, i8 27) nounwind readnone
130  ret <4 x i32> %res
131}
132declare <4 x i32> @llvm.x86.sse2.pshuf.d(<4 x i32>, i8) nounwind readnone
133
134define <8 x i16> @test_x86_sse2_pshufl_w(<8 x i16> %a) {
135; CHECK-LABEL: test_x86_sse2_pshufl_w:
136; CHECK:       ## BB#0: ## %entry
137; CHECK-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,2,1,0,4,5,6,7]
138; CHECK-NEXT:    retl
139entry:
140  %res = call <8 x i16> @llvm.x86.sse2.pshufl.w(<8 x i16> %a, i8 27) nounwind readnone
141  ret <8 x i16> %res
142}
143declare <8 x i16> @llvm.x86.sse2.pshufl.w(<8 x i16>, i8) nounwind readnone
144
145define <8 x i16> @test_x86_sse2_pshufh_w(<8 x i16> %a) {
146; CHECK-LABEL: test_x86_sse2_pshufh_w:
147; CHECK:       ## BB#0: ## %entry
148; CHECK-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,6,5,4]
149; CHECK-NEXT:    retl
150entry:
151  %res = call <8 x i16> @llvm.x86.sse2.pshufh.w(<8 x i16> %a, i8 27) nounwind readnone
152  ret <8 x i16> %res
153}
154declare <8 x i16> @llvm.x86.sse2.pshufh.w(<8 x i16>, i8) nounwind readnone
155
156define <16 x i8> @max_epu8(<16 x i8> %a0, <16 x i8> %a1) {
157; CHECK-LABEL: max_epu8:
158; CHECK:       ## BB#0:
159; CHECK-NEXT:    pmaxub %xmm1, %xmm0
160; CHECK-NEXT:    retl
161;
162  %res = call <16 x i8> @llvm.x86.sse2.pmaxu.b(<16 x i8> %a0, <16 x i8> %a1)
163  ret <16 x i8> %res
164}
165declare <16 x i8> @llvm.x86.sse2.pmaxu.b(<16 x i8>, <16 x i8>) nounwind readnone
166
167define <16 x i8> @min_epu8(<16 x i8> %a0, <16 x i8> %a1) {
168; CHECK-LABEL: min_epu8:
169; CHECK:       ## BB#0:
170; CHECK-NEXT:    pminub %xmm1, %xmm0
171; CHECK-NEXT:    retl
172;
173  %res = call <16 x i8> @llvm.x86.sse2.pminu.b(<16 x i8> %a0, <16 x i8> %a1)
174  ret <16 x i8> %res
175}
176declare <16 x i8> @llvm.x86.sse2.pminu.b(<16 x i8>, <16 x i8>) nounwind readnone
177
178define <8 x i16> @max_epi16(<8 x i16> %a0, <8 x i16> %a1) {
179; CHECK-LABEL: max_epi16:
180; CHECK:       ## BB#0:
181; CHECK-NEXT:    pmaxsw %xmm1, %xmm0
182; CHECK-NEXT:    retl
183;
184  %res = call <8 x i16> @llvm.x86.sse2.pmaxs.w(<8 x i16> %a0, <8 x i16> %a1)
185  ret <8 x i16> %res
186}
187declare <8 x i16> @llvm.x86.sse2.pmaxs.w(<8 x i16>, <8 x i16>) nounwind readnone
188
189define <8 x i16> @min_epi16(<8 x i16> %a0, <8 x i16> %a1) {
190; CHECK-LABEL: min_epi16:
191; CHECK:       ## BB#0:
192; CHECK-NEXT:    pminsw %xmm1, %xmm0
193; CHECK-NEXT:    retl
194;
195  %res = call <8 x i16> @llvm.x86.sse2.pmins.w(<8 x i16> %a0, <8 x i16> %a1)
196  ret <8 x i16> %res
197}
198declare <8 x i16> @llvm.x86.sse2.pmins.w(<8 x i16>, <8 x i16>) nounwind readnone
199
200