xref: /llvm-project/llvm/test/CodeGen/VE/VELIntrinsics/vpcnt.ll (revision 926059214193e3fcab20ee43a1fa734a8f20210e) 
1; RUN: llc < %s -mtriple=ve -mattr=+vpu | FileCheck %s
2
3;;; Test vector population count intrinsic instructions
4;;;
5;;; Note:
6;;;   We test VPCNT*vl, VPCNT*vl_v, VPCNT*vml_v, PVPCNT*vl, PVPCNT*vl_v, PVPCNT*vml_v instructions.
7
8; Function Attrs: nounwind readnone
9define fastcc <256 x double> @vpcnt_vvl(<256 x double> %0) {
10; CHECK-LABEL: vpcnt_vvl:
11; CHECK:       # %bb.0:
12; CHECK-NEXT:    lea %s0, 256
13; CHECK-NEXT:    lvl %s0
14; CHECK-NEXT:    vpcnt %v0, %v0
15; CHECK-NEXT:    b.l.t (, %s10)
16  %2 = tail call fast <256 x double> @llvm.ve.vl.vpcnt.vvl(<256 x double> %0, i32 256)
17  ret <256 x double> %2
18}
19
20; Function Attrs: nounwind readnone
21declare <256 x double> @llvm.ve.vl.vpcnt.vvl(<256 x double>, i32)
22
23; Function Attrs: nounwind readnone
24define fastcc <256 x double> @vpcnt_vvvl(<256 x double> %0, <256 x double> %1) {
25; CHECK-LABEL: vpcnt_vvvl:
26; CHECK:       # %bb.0:
27; CHECK-NEXT:    lea %s0, 128
28; CHECK-NEXT:    lvl %s0
29; CHECK-NEXT:    vpcnt %v1, %v0
30; CHECK-NEXT:    lea %s16, 256
31; CHECK-NEXT:    lvl %s16
32; CHECK-NEXT:    vor %v0, (0)1, %v1
33; CHECK-NEXT:    b.l.t (, %s10)
34  %3 = tail call fast <256 x double> @llvm.ve.vl.vpcnt.vvvl(<256 x double> %0, <256 x double> %1, i32 128)
35  ret <256 x double> %3
36}
37
38; Function Attrs: nounwind readnone
39declare <256 x double> @llvm.ve.vl.vpcnt.vvvl(<256 x double>, <256 x double>, i32)
40
41; Function Attrs: nounwind readnone
42define fastcc <256 x double> @vpcnt_vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2) {
43; CHECK-LABEL: vpcnt_vvmvl:
44; CHECK:       # %bb.0:
45; CHECK-NEXT:    lea %s0, 128
46; CHECK-NEXT:    lvl %s0
47; CHECK-NEXT:    vpcnt %v1, %v0, %vm1
48; CHECK-NEXT:    lea %s16, 256
49; CHECK-NEXT:    lvl %s16
50; CHECK-NEXT:    vor %v0, (0)1, %v1
51; CHECK-NEXT:    b.l.t (, %s10)
52  %4 = tail call fast <256 x double> @llvm.ve.vl.vpcnt.vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2, i32 128)
53  ret <256 x double> %4
54}
55
56; Function Attrs: nounwind readnone
57declare <256 x double> @llvm.ve.vl.vpcnt.vvmvl(<256 x double>, <256 x i1>, <256 x double>, i32)
58
59; Function Attrs: nounwind readnone
60define fastcc <256 x double> @pvpcntlo_vvl(<256 x double> %0) {
61; CHECK-LABEL: pvpcntlo_vvl:
62; CHECK:       # %bb.0:
63; CHECK-NEXT:    lea %s0, 256
64; CHECK-NEXT:    lvl %s0
65; CHECK-NEXT:    pvpcnt.lo %v0, %v0
66; CHECK-NEXT:    b.l.t (, %s10)
67  %2 = tail call fast <256 x double> @llvm.ve.vl.pvpcntlo.vvl(<256 x double> %0, i32 256)
68  ret <256 x double> %2
69}
70
71; Function Attrs: nounwind readnone
72declare <256 x double> @llvm.ve.vl.pvpcntlo.vvl(<256 x double>, i32)
73
74; Function Attrs: nounwind readnone
75define fastcc <256 x double> @pvpcntlo_vvvl(<256 x double> %0, <256 x double> %1) {
76; CHECK-LABEL: pvpcntlo_vvvl:
77; CHECK:       # %bb.0:
78; CHECK-NEXT:    lea %s0, 128
79; CHECK-NEXT:    lvl %s0
80; CHECK-NEXT:    pvpcnt.lo %v1, %v0
81; CHECK-NEXT:    lea %s16, 256
82; CHECK-NEXT:    lvl %s16
83; CHECK-NEXT:    vor %v0, (0)1, %v1
84; CHECK-NEXT:    b.l.t (, %s10)
85  %3 = tail call fast <256 x double> @llvm.ve.vl.pvpcntlo.vvvl(<256 x double> %0, <256 x double> %1, i32 128)
86  ret <256 x double> %3
87}
88
89; Function Attrs: nounwind readnone
90declare <256 x double> @llvm.ve.vl.pvpcntlo.vvvl(<256 x double>, <256 x double>, i32)
91
92; Function Attrs: nounwind readnone
93define fastcc <256 x double> @pvpcntlo_vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2) {
94; CHECK-LABEL: pvpcntlo_vvmvl:
95; CHECK:       # %bb.0:
96; CHECK-NEXT:    lea %s0, 128
97; CHECK-NEXT:    lvl %s0
98; CHECK-NEXT:    pvpcnt.lo %v1, %v0, %vm1
99; CHECK-NEXT:    lea %s16, 256
100; CHECK-NEXT:    lvl %s16
101; CHECK-NEXT:    vor %v0, (0)1, %v1
102; CHECK-NEXT:    b.l.t (, %s10)
103  %4 = tail call fast <256 x double> @llvm.ve.vl.pvpcntlo.vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2, i32 128)
104  ret <256 x double> %4
105}
106
107; Function Attrs: nounwind readnone
108declare <256 x double> @llvm.ve.vl.pvpcntlo.vvmvl(<256 x double>, <256 x i1>, <256 x double>, i32)
109
110; Function Attrs: nounwind readnone
111define fastcc <256 x double> @pvpcntup_vvl(<256 x double> %0) {
112; CHECK-LABEL: pvpcntup_vvl:
113; CHECK:       # %bb.0:
114; CHECK-NEXT:    lea %s0, 256
115; CHECK-NEXT:    lvl %s0
116; CHECK-NEXT:    pvpcnt.up %v0, %v0
117; CHECK-NEXT:    b.l.t (, %s10)
118  %2 = tail call fast <256 x double> @llvm.ve.vl.pvpcntup.vvl(<256 x double> %0, i32 256)
119  ret <256 x double> %2
120}
121
122; Function Attrs: nounwind readnone
123declare <256 x double> @llvm.ve.vl.pvpcntup.vvl(<256 x double>, i32)
124
125; Function Attrs: nounwind readnone
126define fastcc <256 x double> @pvpcntup_vvvl(<256 x double> %0, <256 x double> %1) {
127; CHECK-LABEL: pvpcntup_vvvl:
128; CHECK:       # %bb.0:
129; CHECK-NEXT:    lea %s0, 128
130; CHECK-NEXT:    lvl %s0
131; CHECK-NEXT:    pvpcnt.up %v1, %v0
132; CHECK-NEXT:    lea %s16, 256
133; CHECK-NEXT:    lvl %s16
134; CHECK-NEXT:    vor %v0, (0)1, %v1
135; CHECK-NEXT:    b.l.t (, %s10)
136  %3 = tail call fast <256 x double> @llvm.ve.vl.pvpcntup.vvvl(<256 x double> %0, <256 x double> %1, i32 128)
137  ret <256 x double> %3
138}
139
140; Function Attrs: nounwind readnone
141declare <256 x double> @llvm.ve.vl.pvpcntup.vvvl(<256 x double>, <256 x double>, i32)
142
143; Function Attrs: nounwind readnone
144define fastcc <256 x double> @pvpcntup_vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2) {
145; CHECK-LABEL: pvpcntup_vvmvl:
146; CHECK:       # %bb.0:
147; CHECK-NEXT:    lea %s0, 128
148; CHECK-NEXT:    lvl %s0
149; CHECK-NEXT:    pvpcnt.up %v1, %v0, %vm1
150; CHECK-NEXT:    lea %s16, 256
151; CHECK-NEXT:    lvl %s16
152; CHECK-NEXT:    vor %v0, (0)1, %v1
153; CHECK-NEXT:    b.l.t (, %s10)
154  %4 = tail call fast <256 x double> @llvm.ve.vl.pvpcntup.vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2, i32 128)
155  ret <256 x double> %4
156}
157
158; Function Attrs: nounwind readnone
159declare <256 x double> @llvm.ve.vl.pvpcntup.vvmvl(<256 x double>, <256 x i1>, <256 x double>, i32)
160
161; Function Attrs: nounwind readnone
162define fastcc <256 x double> @pvpcnt_vvl(<256 x double> %0) {
163; CHECK-LABEL: pvpcnt_vvl:
164; CHECK:       # %bb.0:
165; CHECK-NEXT:    lea %s0, 256
166; CHECK-NEXT:    lvl %s0
167; CHECK-NEXT:    pvpcnt %v0, %v0
168; CHECK-NEXT:    b.l.t (, %s10)
169  %2 = tail call fast <256 x double> @llvm.ve.vl.pvpcnt.vvl(<256 x double> %0, i32 256)
170  ret <256 x double> %2
171}
172
173; Function Attrs: nounwind readnone
174declare <256 x double> @llvm.ve.vl.pvpcnt.vvl(<256 x double>, i32)
175
176; Function Attrs: nounwind readnone
177define fastcc <256 x double> @pvpcnt_vvvl(<256 x double> %0, <256 x double> %1) {
178; CHECK-LABEL: pvpcnt_vvvl:
179; CHECK:       # %bb.0:
180; CHECK-NEXT:    lea %s0, 128
181; CHECK-NEXT:    lvl %s0
182; CHECK-NEXT:    pvpcnt %v1, %v0
183; CHECK-NEXT:    lea %s16, 256
184; CHECK-NEXT:    lvl %s16
185; CHECK-NEXT:    vor %v0, (0)1, %v1
186; CHECK-NEXT:    b.l.t (, %s10)
187  %3 = tail call fast <256 x double> @llvm.ve.vl.pvpcnt.vvvl(<256 x double> %0, <256 x double> %1, i32 128)
188  ret <256 x double> %3
189}
190
191; Function Attrs: nounwind readnone
192declare <256 x double> @llvm.ve.vl.pvpcnt.vvvl(<256 x double>, <256 x double>, i32)
193
194; Function Attrs: nounwind readnone
195define fastcc <256 x double> @pvpcnt_vvMvl(<256 x double> %0, <512 x i1> %1, <256 x double> %2) {
196; CHECK-LABEL: pvpcnt_vvMvl:
197; CHECK:       # %bb.0:
198; CHECK-NEXT:    lea %s0, 128
199; CHECK-NEXT:    lvl %s0
200; CHECK-NEXT:    pvpcnt %v1, %v0, %vm2
201; CHECK-NEXT:    lea %s16, 256
202; CHECK-NEXT:    lvl %s16
203; CHECK-NEXT:    vor %v0, (0)1, %v1
204; CHECK-NEXT:    b.l.t (, %s10)
205  %4 = tail call fast <256 x double> @llvm.ve.vl.pvpcnt.vvMvl(<256 x double> %0, <512 x i1> %1, <256 x double> %2, i32 128)
206  ret <256 x double> %4
207}
208
209; Function Attrs: nounwind readnone
210declare <256 x double> @llvm.ve.vl.pvpcnt.vvMvl(<256 x double>, <512 x i1>, <256 x double>, i32)
211