Skip to content

Commit 015b381

Browse files
goldsteinngoldsteinn
committed
Add tests for shuffle as shift/rotate; NFC
Differential Revision: https://reviews.llvm.org/D143857
1 parent 5767497 commit 015b381

File tree

1 file changed

+242
-0
lines changed

1 file changed

+242
-0
lines changed

‎llvm/test/CodeGen/X86/shuffle-as-shifts.ll

Lines changed: 242 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,242 @@
1+
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2+
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=skylake-avx512 | FileCheck %s --check-prefixes=CHECK,CHECK-SKX
3+
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=icelake-server | FileCheck %s --check-prefixes=CHECK,CHECK-ICX
4+
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=CHECK,CHECK-V4
5+
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=znver4 | FileCheck %s --check-prefixes=CHECK,CHECK-ZNVER4
6+
7+
8+
define <4 x i32> @shuf_rot_v4i32_1032(<4 x i32> %x) {
9+
; CHECK-LABEL: shuf_rot_v4i32_1032:
10+
; CHECK: # %bb.0:
11+
; CHECK-NEXT: vpaddd %xmm0, %xmm0, %xmm0
12+
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2]
13+
; CHECK-NEXT: retq
14+
%x1 = add <4 x i32> %x, %x
15+
%r = shufflevector <4 x i32> %x1, <4 x i32> zeroinitializer, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
16+
ret <4 x i32> %r
17+
}
18+
19+
define <8 x i32> @shuf_rot_v8i32_10325476(<8 x i32> %x) {
20+
; CHECK-LABEL: shuf_rot_v8i32_10325476:
21+
; CHECK: # %bb.0:
22+
; CHECK-NEXT: vpaddd %ymm0, %ymm0, %ymm0
23+
; CHECK-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,0,3,2,5,4,7,6]
24+
; CHECK-NEXT: retq
25+
%x1 = add <8 x i32> %x, %x
26+
%r = shufflevector <8 x i32> %x1, <8 x i32> zeroinitializer, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
27+
ret <8 x i32> %r
28+
}
29+
30+
define <16 x i32> @shuf_rot_v16i32_1032547698111013121514(<16 x i32> %x) {
31+
; CHECK-LABEL: shuf_rot_v16i32_1032547698111013121514:
32+
; CHECK: # %bb.0:
33+
; CHECK-NEXT: vpaddd %zmm0, %zmm0, %zmm0
34+
; CHECK-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
35+
; CHECK-NEXT: retq
36+
%x1 = add <16 x i32> %x, %x
37+
%r = shufflevector <16 x i32> %x1, <16 x i32> zeroinitializer, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
38+
ret <16 x i32> %r
39+
}
40+
41+
define <8 x i16> @shuf_rot_v8i16_10325476(<8 x i16> %x) {
42+
; CHECK-LABEL: shuf_rot_v8i16_10325476:
43+
; CHECK: # %bb.0:
44+
; CHECK-NEXT: vpaddw %xmm0, %xmm0, %xmm0
45+
; CHECK-NEXT: vprold $16, %xmm0, %xmm0
46+
; CHECK-NEXT: retq
47+
%x1 = add <8 x i16> %x, %x
48+
%r = shufflevector <8 x i16> %x1, <8 x i16> zeroinitializer, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
49+
ret <8 x i16> %r
50+
}
51+
52+
define <16 x i16> @shuf_rot_v16i16_1032547698111013121514(<16 x i16> %x) {
53+
; CHECK-LABEL: shuf_rot_v16i16_1032547698111013121514:
54+
; CHECK: # %bb.0:
55+
; CHECK-NEXT: vpaddw %ymm0, %ymm0, %ymm0
56+
; CHECK-NEXT: vprold $16, %ymm0, %ymm0
57+
; CHECK-NEXT: retq
58+
%x1 = add <16 x i16> %x, %x
59+
%r = shufflevector <16 x i16> %x1, <16 x i16> zeroinitializer, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
60+
ret <16 x i16> %r
61+
}
62+
63+
define <32 x i16> @shuf_rot_v32i16_1234056749101181314151217181916212223202527272429303128(<32 x i16> %x) {
64+
; CHECK-LABEL: shuf_rot_v32i16_1234056749101181314151217181916212223202527272429303128:
65+
; CHECK: # %bb.0:
66+
; CHECK-NEXT: vpaddw %zmm0, %zmm0, %zmm0
67+
; CHECK-NEXT: vprolq $48, %zmm0, %zmm0
68+
; CHECK-NEXT: retq
69+
%x1 = add <32 x i16> %x, %x
70+
%r = shufflevector <32 x i16> %x1, <32 x i16> zeroinitializer, <32 x i32> <i32 1,i32 2,i32 3,i32 0,i32 5,i32 6,i32 7,i32 4,i32 9,i32 10,i32 11,i32 8,i32 13,i32 14,i32 15,i32 12,i32 17,i32 18,i32 19,i32 16,i32 21,i32 22,i32 23,i32 20,i32 25,i32 26,i32 27,i32 24,i32 29,i32 30,i32 31,i32 28>
71+
ret <32 x i16> %r
72+
}
73+
74+
define <16 x i8> @shuf_rot_v16i8_2301674510118914151213(<16 x i8> %x) {
75+
; CHECK-LABEL: shuf_rot_v16i8_2301674510118914151213:
76+
; CHECK: # %bb.0:
77+
; CHECK-NEXT: vpaddb %xmm0, %xmm0, %xmm0
78+
; CHECK-NEXT: vprold $16, %xmm0, %xmm0
79+
; CHECK-NEXT: retq
80+
%x1 = add <16 x i8> %x, %x
81+
%r = shufflevector <16 x i8> %x1, <16 x i8> zeroinitializer, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9, i32 14, i32 15, i32 12, i32 13>
82+
ret <16 x i8> %r
83+
}
84+
85+
define <32 x i8> @shuf_rot_v32i8_230167451011891415121318191617222320212627242530312829(<32 x i8> %x) {
86+
; CHECK-LABEL: shuf_rot_v32i8_230167451011891415121318191617222320212627242530312829:
87+
; CHECK: # %bb.0:
88+
; CHECK-NEXT: vpaddb %ymm0, %ymm0, %ymm0
89+
; CHECK-NEXT: vprold $16, %ymm0, %ymm0
90+
; CHECK-NEXT: retq
91+
%x1 = add <32 x i8> %x, %x
92+
%r = shufflevector <32 x i8> %x1, <32 x i8> zeroinitializer, <32 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9, i32 14, i32 15, i32 12, i32 13, i32 18, i32 19, i32 16, i32 17, i32 22, i32 23, i32 20, i32 21, i32 26, i32 27, i32 24, i32 25, i32 30, i32 31, i32 28, i32 29>
93+
ret <32 x i8> %r
94+
}
95+
96+
define <64 x i8> @shuf_rot_v64i8_3012745611891015121314191617182320212227242526312829303532333439363738434041424744454651484950555253545956575863606162(<64 x i8> %x) {
97+
; CHECK-LABEL: shuf_rot_v64i8_3012745611891015121314191617182320212227242526312829303532333439363738434041424744454651484950555253545956575863606162:
98+
; CHECK: # %bb.0:
99+
; CHECK-NEXT: vpaddb %zmm0, %zmm0, %zmm0
100+
; CHECK-NEXT: vprold $8, %zmm0, %zmm0
101+
; CHECK-NEXT: retq
102+
%x1 = add <64 x i8> %x, %x
103+
%r = shufflevector <64 x i8> %x1, <64 x i8> zeroinitializer, <64 x i32> <i32 3,i32 0,i32 1,i32 2,i32 7,i32 4,i32 5,i32 6,i32 11,i32 8,i32 9,i32 10,i32 15,i32 12,i32 13,i32 14,i32 19,i32 16,i32 17,i32 18,i32 23,i32 20,i32 21,i32 22,i32 27,i32 24,i32 25,i32 26,i32 31,i32 28,i32 29,i32 30,i32 35,i32 32,i32 33,i32 34,i32 39,i32 36,i32 37,i32 38,i32 43,i32 40,i32 41,i32 42,i32 47,i32 44,i32 45,i32 46,i32 51,i32 48,i32 49,i32 50,i32 55,i32 52,i32 53,i32 54,i32 59,i32 56,i32 57,i32 58,i32 63,i32 60,i32 61,i32 62>
104+
ret <64 x i8> %r
105+
}
106+
107+
define <4 x i32> @shuf_shr_v4i32_1U3U(<4 x i32> %x) {
108+
; CHECK-LABEL: shuf_shr_v4i32_1U3U:
109+
; CHECK: # %bb.0:
110+
; CHECK-NEXT: vpaddd %xmm0, %xmm0, %xmm0
111+
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
112+
; CHECK-NEXT: retq
113+
%x1 = add <4 x i32> %x, %x
114+
%r = shufflevector <4 x i32> %x1, <4 x i32> zeroinitializer, <4 x i32> <i32 1, i32 undef, i32 3, i32 undef>
115+
ret <4 x i32> %r
116+
}
117+
118+
define <8 x i32> @shuf_shr_v8i32_1U3U5U7U(<8 x i32> %x) {
119+
; CHECK-LABEL: shuf_shr_v8i32_1U3U5U7U:
120+
; CHECK: # %bb.0:
121+
; CHECK-NEXT: vpaddd %ymm0, %ymm0, %ymm0
122+
; CHECK-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,1,3,3,5,5,7,7]
123+
; CHECK-NEXT: retq
124+
%x1 = add <8 x i32> %x, %x
125+
%r = shufflevector <8 x i32> %x1, <8 x i32> zeroinitializer, <8 x i32> <i32 1, i32 undef, i32 3, i32 undef, i32 5, i32 undef, i32 7, i32 undef>
126+
ret <8 x i32> %r
127+
}
128+
129+
define <16 x i32> @shuf_shr_v16i32_U3U5U7U9U11U13U15(<16 x i32> %x) {
130+
; CHECK-LABEL: shuf_shr_v16i32_U3U5U7U9U11U13U15:
131+
; CHECK: # %bb.0:
132+
; CHECK-NEXT: vpaddd %zmm0, %zmm0, %zmm0
133+
; CHECK-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
134+
; CHECK-NEXT: retq
135+
%x1 = add <16 x i32> %x, %x
136+
%r = shufflevector <16 x i32> %x1, <16 x i32> zeroinitializer, <16 x i32> <i32 1, i32 undef, i32 3, i32 undef, i32 5, i32 undef, i32 7, i32 undef, i32 9, i32 undef, i32 11, i32 undef, i32 13, i32 undef, i32 15, i32 undef>
137+
ret <16 x i32> %r
138+
}
139+
140+
define <8 x i16> @shuf_shr_v8i16_123U567U(<8 x i16> %x) {
141+
; CHECK-LABEL: shuf_shr_v8i16_123U567U:
142+
; CHECK: # %bb.0:
143+
; CHECK-NEXT: vpaddw %xmm0, %xmm0, %xmm0
144+
; CHECK-NEXT: vpsrlq $16, %xmm0, %xmm0
145+
; CHECK-NEXT: retq
146+
%x1 = add <8 x i16> %x, %x
147+
%r = shufflevector <8 x i16> %x1, <8 x i16> zeroinitializer, <8 x i32> <i32 1, i32 2, i32 3, i32 undef, i32 5, i32 6, i32 7, i32 undef>
148+
ret <8 x i16> %r
149+
}
150+
151+
define <32 x i16> @shuf_shr_v32i16_1U3U5U7U9U11U13U15U17U19U21U23U25U27U29U31U(<32 x i16> %x) {
152+
; CHECK-LABEL: shuf_shr_v32i16_1U3U5U7U9U11U13U15U17U19U21U23U25U27U29U31U:
153+
; CHECK: # %bb.0:
154+
; CHECK-NEXT: vpaddw %zmm0, %zmm0, %zmm0
155+
; CHECK-NEXT: vpsrld $16, %zmm0, %zmm0
156+
; CHECK-NEXT: retq
157+
%x1 = add <32 x i16> %x, %x
158+
%r = shufflevector <32 x i16> %x1, <32 x i16> zeroinitializer, <32 x i32> <i32 1, i32 undef, i32 3, i32 undef, i32 5, i32 undef, i32 7, i32 undef, i32 9, i32 undef, i32 11, i32 undef, i32 13, i32 undef, i32 15, i32 undef, i32 17, i32 undef, i32 19, i32 undef, i32 21, i32 undef, i32 23, i32 undef, i32 25, i32 undef, i32 27, i32 undef, i32 29, i32 undef, i32 31, i32 undef>
159+
ret <32 x i16> %r
160+
}
161+
162+
define <32 x i8> @shuf_shr_v32i8_1U3U5U7U9U11U13U15U17U19U21U23U25U27U29U31U(<32 x i8> %x) {
163+
; CHECK-LABEL: shuf_shr_v32i8_1U3U5U7U9U11U13U15U17U19U21U23U25U27U29U31U:
164+
; CHECK: # %bb.0:
165+
; CHECK-NEXT: vpaddb %ymm0, %ymm0, %ymm0
166+
; CHECK-NEXT: vpsrlw $8, %ymm0, %ymm0
167+
; CHECK-NEXT: retq
168+
%x1 = add <32 x i8> %x, %x
169+
%r = shufflevector <32 x i8> %x1, <32 x i8> zeroinitializer, <32 x i32> <i32 1, i32 undef, i32 3, i32 undef, i32 5, i32 undef, i32 7, i32 undef, i32 9, i32 undef, i32 11, i32 undef, i32 13, i32 undef, i32 15, i32 undef, i32 17, i32 undef, i32 19, i32 undef, i32 21, i32 undef, i32 23, i32 undef, i32 25, i32 undef, i32 27, i32 undef, i32 29, i32 undef, i32 31, i32 undef>
170+
ret <32 x i8> %r
171+
}
172+
173+
define <4 x i32> @shuf_shl_v4i32_U0U2(<4 x i32> %x) {
174+
; CHECK-LABEL: shuf_shl_v4i32_U0U2:
175+
; CHECK: # %bb.0:
176+
; CHECK-NEXT: vpaddd %xmm0, %xmm0, %xmm0
177+
; CHECK-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,2,2]
178+
; CHECK-NEXT: retq
179+
%x1 = add <4 x i32> %x, %x
180+
%r = shufflevector <4 x i32> %x1, <4 x i32> zeroinitializer, <4 x i32> <i32 undef, i32 0, i32 undef, i32 2>
181+
ret <4 x i32> %r
182+
}
183+
184+
define <8 x i32> @shuf_shl_v8i32_U0U2U4U6(<8 x i32> %x) {
185+
; CHECK-LABEL: shuf_shl_v8i32_U0U2U4U6:
186+
; CHECK: # %bb.0:
187+
; CHECK-NEXT: vpaddd %ymm0, %ymm0, %ymm0
188+
; CHECK-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,0,2,2,4,4,6,6]
189+
; CHECK-NEXT: retq
190+
%x1 = add <8 x i32> %x, %x
191+
%r = shufflevector <8 x i32> %x1, <8 x i32> zeroinitializer, <8 x i32> <i32 undef, i32 0, i32 undef, i32 2, i32 undef, i32 4, i32 undef, i32 6>
192+
ret <8 x i32> %r
193+
}
194+
195+
define <16 x i32> @shuf_shl_v16i32_U0U2U4U6U8U10U12U14(<16 x i32> %x) {
196+
; CHECK-LABEL: shuf_shl_v16i32_U0U2U4U6U8U10U12U14:
197+
; CHECK: # %bb.0:
198+
; CHECK-NEXT: vpaddd %zmm0, %zmm0, %zmm0
199+
; CHECK-NEXT: vpshufd {{.*#+}} zmm0 = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
200+
; CHECK-NEXT: retq
201+
%x1 = add <16 x i32> %x, %x
202+
%r = shufflevector <16 x i32> %x1, <16 x i32> zeroinitializer, <16 x i32> <i32 undef, i32 0, i32 undef, i32 2, i32 undef, i32 4, i32 undef, i32 6, i32 undef, i32 8, i32 undef, i32 10, i32 undef, i32 12, i32 undef, i32 14>
203+
ret <16 x i32> %r
204+
}
205+
206+
define <16 x i16> @shuf_shl_v16i16_U0U2U4U6U8U10U12U14(<16 x i16> %x) {
207+
; CHECK-LABEL: shuf_shl_v16i16_U0U2U4U6U8U10U12U14:
208+
; CHECK: # %bb.0:
209+
; CHECK-NEXT: vpaddw %ymm0, %ymm0, %ymm0
210+
; CHECK-NEXT: vpslld $16, %ymm0, %ymm0
211+
; CHECK-NEXT: retq
212+
%x1 = add <16 x i16> %x, %x
213+
%r = shufflevector <16 x i16> %x1, <16 x i16> zeroinitializer, <16 x i32> <i32 undef, i32 0, i32 undef, i32 2, i32 undef, i32 4, i32 undef, i32 6, i32 undef, i32 8, i32 undef, i32 10, i32 undef, i32 12, i32 undef, i32 14>
214+
ret <16 x i16> %r
215+
}
216+
217+
define <16 x i8> @shuf_shl_v16i8_U0U2U4U6U8U10U12U14(<16 x i8> %x) {
218+
; CHECK-LABEL: shuf_shl_v16i8_U0U2U4U6U8U10U12U14:
219+
; CHECK: # %bb.0:
220+
; CHECK-NEXT: vpaddb %xmm0, %xmm0, %xmm0
221+
; CHECK-NEXT: vpsllw $8, %xmm0, %xmm0
222+
; CHECK-NEXT: retq
223+
%x1 = add <16 x i8> %x, %x
224+
%r = shufflevector <16 x i8> %x1, <16 x i8> zeroinitializer, <16 x i32> <i32 undef, i32 0, i32 undef, i32 2, i32 undef, i32 4, i32 undef, i32 6, i32 undef, i32 8, i32 undef, i32 10, i32 undef, i32 12, i32 undef, i32 14>
225+
ret <16 x i8> %r
226+
}
227+
228+
define <64 x i8> @shuf_shl_v64i8_U0U2U4U6U8U10U12U14U16U18U20U22U24U26U28U30U32U34U36U38U40U42U44U46U48U50U52U54U56U58U60U62(<64 x i8> %x) {
229+
; CHECK-LABEL: shuf_shl_v64i8_U0U2U4U6U8U10U12U14U16U18U20U22U24U26U28U30U32U34U36U38U40U42U44U46U48U50U52U54U56U58U60U62:
230+
; CHECK: # %bb.0:
231+
; CHECK-NEXT: vpaddb %zmm0, %zmm0, %zmm0
232+
; CHECK-NEXT: vpsllw $8, %zmm0, %zmm0
233+
; CHECK-NEXT: retq
234+
%x1 = add <64 x i8> %x, %x
235+
%r = shufflevector <64 x i8> %x1, <64 x i8> zeroinitializer, <64 x i32> <i32 undef, i32 0, i32 undef, i32 2, i32 undef, i32 4, i32 undef, i32 6, i32 undef, i32 8, i32 undef, i32 10, i32 undef, i32 12, i32 undef, i32 14, i32 undef, i32 16, i32 undef, i32 18, i32 undef, i32 20, i32 undef, i32 22, i32 undef, i32 24, i32 undef, i32 26, i32 undef, i32 28, i32 undef, i32 30, i32 undef, i32 32, i32 undef, i32 34, i32 undef, i32 36, i32 undef, i32 38, i32 undef, i32 40, i32 undef, i32 42, i32 undef, i32 44, i32 undef, i32 46, i32 undef, i32 48, i32 undef, i32 50, i32 undef, i32 52, i32 undef, i32 54, i32 undef, i32 56, i32 undef, i32 58, i32 undef, i32 60, i32 undef, i32 62>
236+
ret <64 x i8> %r
237+
}
238+
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
239+
; CHECK-ICX: {{.*}}
240+
; CHECK-SKX: {{.*}}
241+
; CHECK-V4: {{.*}}
242+
; CHECK-ZNVER4: {{.*}}

0 commit comments

Comments
 (0)