Skip to content

Commit 81acc84

Browse files
RKSimonRKSimon
committed
[LoopVectorize][X86] amdlibm-calls.ll - add 2/4/8/16 vector widths test checks for fallback to llvm intrinsics
Check for cases where there isn't a amdlib call but it still vectorises the math call
1 parent 115b876 commit 81acc84

File tree

1 file changed

+34
-0
lines changed

1 file changed

+34
-0
lines changed

‎llvm/test/Transforms/LoopVectorize/X86/amdlibm-calls.ll‎

Lines changed: 34 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -80,6 +80,7 @@ define void @sin_f64(ptr nocapture %varray) {
8080
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_sin(<2 x double> [[TMP4:%.*]])
8181
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_sin(<4 x double> [[TMP4:%.*]])
8282
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_sin(<8 x double> [[TMP4:%.*]])
83+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.sin.v16f64(<16 x double> [[TMP4:%.*]])
8384
; CHECK: ret void
8485
;
8586
entry:
@@ -102,6 +103,7 @@ for.end:
102103

103104
define void @sin_f32(ptr nocapture %varray) {
104105
; CHECK-LABEL: @sin_f32(
106+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.sin.v2f32(<2 x float> [[TMP4:%.*]])
105107
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_sinf(<4 x float> [[TMP4:%.*]])
106108
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_sinf(<8 x float> [[TMP4:%.*]])
107109
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_sinf(<16 x float> [[TMP4:%.*]])
@@ -130,6 +132,7 @@ define void @sin_f64_intrinsic(ptr nocapture %varray) {
130132
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_sin(<2 x double> [[TMP4:%.*]])
131133
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_sin(<4 x double> [[TMP4:%.*]])
132134
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_sin(<8 x double> [[TMP4:%.*]])
135+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.sin.v16f64(<16 x double> [[TMP4:%.*]])
133136
; CHECK: ret void
134137
;
135138
entry:
@@ -152,6 +155,7 @@ for.end:
152155

153156
define void @sin_f32_intrinsic(ptr nocapture %varray) {
154157
; CHECK-LABEL: @sin_f32_intrinsic(
158+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.sin.v2f32(<2 x float> [[TMP4:%.*]])
155159
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_sinf(<4 x float> [[TMP4:%.*]])
156160
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_sinf(<8 x float> [[TMP4:%.*]])
157161
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_sinf(<16 x float> [[TMP4:%.*]])
@@ -180,6 +184,7 @@ define void @cos_f64(ptr nocapture %varray) {
180184
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_cos(<2 x double> [[TMP4:%.*]])
181185
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_cos(<4 x double> [[TMP4:%.*]])
182186
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_cos(<8 x double> [[TMP4:%.*]])
187+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.cos.v16f64(<16 x double> [[TMP4:%.*]])
183188
; CHECK: ret void
184189
;
185190
entry:
@@ -202,6 +207,7 @@ for.end:
202207

203208
define void @cos_f32(ptr nocapture %varray) {
204209
; CHECK-LABEL: @cos_f32(
210+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.cos.v2f32(<2 x float> [[TMP4:%.*]])
205211
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_cosf(<4 x float> [[TMP4:%.*]])
206212
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_cosf(<8 x float> [[TMP4:%.*]])
207213
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_cosf(<16 x float> [[TMP4:%.*]])
@@ -230,6 +236,7 @@ define void @cos_f64_intrinsic(ptr nocapture %varray) {
230236
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_cos(<2 x double> [[TMP4:%.*]])
231237
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_cos(<4 x double> [[TMP4:%.*]])
232238
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_cos(<8 x double> [[TMP4:%.*]])
239+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.cos.v16f64(<16 x double> [[TMP4:%.*]])
233240
; CHECK: ret void
234241
;
235242
entry:
@@ -252,6 +259,7 @@ for.end:
252259

253260
define void @cos_f32_intrinsic(ptr nocapture %varray) {
254261
; CHECK-LABEL: @cos_f32_intrinsic(
262+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.cos.v2f32(<2 x float> [[TMP4:%.*]])
255263
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_cosf(<4 x float> [[TMP4:%.*]])
256264
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_cosf(<8 x float> [[TMP4:%.*]])
257265
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_cosf(<16 x float> [[TMP4:%.*]])
@@ -280,6 +288,7 @@ define void @tan_f64(ptr nocapture %varray) {
280288
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_tan(<2 x double> [[TMP4:%.*]])
281289
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_tan(<4 x double> [[TMP4:%.*]])
282290
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_tan(<8 x double> [[TMP4:%.*]])
291+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.tan.v16f64(<16 x double> [[TMP4:%.*]])
283292
; CHECK: ret void
284293
;
285294
entry:
@@ -302,6 +311,7 @@ for.end:
302311

303312
define void @tan_f32(ptr nocapture %varray) {
304313
; CHECK-LABEL: @tan_f32(
314+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.tan.v2f32(<2 x float> [[TMP4:%.*]])
305315
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_tanf(<4 x float> [[TMP4:%.*]])
306316
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_tanf(<8 x float> [[TMP4:%.*]])
307317
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_tanf(<16 x float> [[TMP4:%.*]])
@@ -330,6 +340,7 @@ define void @tan_f64_intrinsic(ptr nocapture %varray) {
330340
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_tan(<2 x double> [[TMP4:%.*]])
331341
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_tan(<4 x double> [[TMP4:%.*]])
332342
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_tan(<8 x double> [[TMP4:%.*]])
343+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.tan.v16f64(<16 x double> [[TMP4:%.*]])
333344
; CHECK: ret void
334345
;
335346
entry:
@@ -352,6 +363,7 @@ for.end:
352363

353364
define void @tan_f32_intrinsic(ptr nocapture %varray) {
354365
; CHECK-LABEL: @tan_f32_intrinsic(
366+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.tan.v2f32(<2 x float> [[TMP4:%.*]])
355367
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_tanf(<4 x float> [[TMP4:%.*]])
356368
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_tanf(<8 x float> [[TMP4:%.*]])
357369
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_tanf(<16 x float> [[TMP4:%.*]])
@@ -770,6 +782,7 @@ define void @pow_f64(ptr nocapture %varray, ptr nocapture readonly %exp) {
770782
; CHECK-VF2: [[TMP8:%.*]] = call <2 x double> @amd_vrd2_pow(<2 x double> [[TMP4:%.*]], <2 x double> [[WIDE_LOAD:%.*]])
771783
; CHECK-VF4: [[TMP8:%.*]] = call <4 x double> @amd_vrd4_pow(<4 x double> [[TMP4:%.*]], <4 x double> [[WIDE_LOAD:%.*]])
772784
; CHECK-VF8: [[TMP8:%.*]] = call <8 x double> @amd_vrd8_pow(<8 x double> [[TMP4:%.*]], <8 x double> [[WIDE_LOAD:%.*]])
785+
; CHECK-VF16: [[TMP8:%.*]] = call <16 x double> @llvm.pow.v16f64(<16 x double> [[TMP4:%.*]], <16 x double> [[WIDE_LOAD:%.*]])
773786
; CHECK: ret void
774787
;
775788
entry:
@@ -797,6 +810,7 @@ define void @pow_f64_intrinsic(ptr nocapture %varray, ptr nocapture readonly %ex
797810
; CHECK-VF2: [[TMP8:%.*]] = call <2 x double> @amd_vrd2_pow(<2 x double> [[TMP4:%.*]], <2 x double> [[WIDE_LOAD:%.*]])
798811
; CHECK-VF4: [[TMP8:%.*]] = call <4 x double> @amd_vrd4_pow(<4 x double> [[TMP4:%.*]], <4 x double> [[WIDE_LOAD:%.*]])
799812
; CHECK-VF8: [[TMP8:%.*]] = call <8 x double> @amd_vrd8_pow(<8 x double> [[TMP4:%.*]], <8 x double> [[WIDE_LOAD:%.*]])
813+
; CHECK-VF16: [[TMP8:%.*]] = call <16 x double> @llvm.pow.v16f64(<16 x double> [[TMP4:%.*]], <16 x double> [[WIDE_LOAD:%.*]])
800814
; CHECK: ret void
801815
;
802816
entry:
@@ -821,6 +835,7 @@ for.end:
821835

822836
define void @pow_f32(ptr nocapture %varray, ptr nocapture readonly %exp) {
823837
; CHECK-LABEL: @pow_f32(
838+
; CHECK-VF2: [[TMP8:%.*]] = call <2 x float> @llvm.pow.v2f32(<2 x float> [[TMP4:%.*]], <2 x float> [[WIDE_LOAD:%.*]])
824839
; CHECK-VF4: [[TMP8:%.*]] = call <4 x float> @amd_vrs4_powf(<4 x float> [[TMP4:%.*]], <4 x float> [[WIDE_LOAD:%.*]])
825840
; CHECK-VF8: [[TMP8:%.*]] = call <8 x float> @amd_vrs8_powf(<8 x float> [[TMP4:%.*]], <8 x float> [[WIDE_LOAD:%.*]])
826841
; CHECK-VF16: [[TMP8:%.*]] = call <16 x float> @amd_vrs16_powf(<16 x float> [[TMP4:%.*]], <16 x float> [[WIDE_LOAD:%.*]])
@@ -848,6 +863,7 @@ for.end:
848863

849864
define void @pow_f32_intrinsic(ptr nocapture %varray, ptr nocapture readonly %exp) {
850865
; CHECK-LABEL: @pow_f32_intrinsic(
866+
; CHECK-VF2: [[TMP8:%.*]] = call <2 x float> @llvm.pow.v2f32(<2 x float> [[TMP4:%.*]], <2 x float> [[WIDE_LOAD:%.*]])
851867
; CHECK-VF4: [[TMP8:%.*]] = call <4 x float> @amd_vrs4_powf(<4 x float> [[TMP4:%.*]], <4 x float> [[WIDE_LOAD:%.*]])
852868
; CHECK-VF8: [[TMP8:%.*]] = call <8 x float> @amd_vrs8_powf(<8 x float> [[TMP4:%.*]], <8 x float> [[WIDE_LOAD:%.*]])
853869
; CHECK-VF16: [[TMP8:%.*]] = call <16 x float> @amd_vrs16_powf(<16 x float> [[TMP4:%.*]], <16 x float> [[WIDE_LOAD:%.*]])
@@ -878,6 +894,7 @@ define void @exp_f64(ptr nocapture %varray) {
878894
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_exp(<2 x double> [[TMP4:%.*]])
879895
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_exp(<4 x double> [[TMP4:%.*]])
880896
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_exp(<8 x double> [[TMP4:%.*]])
897+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.exp.v16f64(<16 x double> [[TMP4:%.*]])
881898
; CHECK: ret void
882899
;
883900
entry:
@@ -900,6 +917,7 @@ for.end:
900917

901918
define void @exp_f32(ptr nocapture %varray) {
902919
; CHECK-LABEL: @exp_f32(
920+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.exp.v2f32(<2 x float> [[TMP4:%.*]])
903921
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_expf(<4 x float> [[TMP4:%.*]])
904922
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_expf(<8 x float> [[TMP4:%.*]])
905923
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_expf(<16 x float> [[TMP4:%.*]])
@@ -928,6 +946,7 @@ define void @exp_f64_intrinsic(ptr nocapture %varray) {
928946
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_exp(<2 x double> [[TMP4:%.*]])
929947
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_exp(<4 x double> [[TMP4:%.*]])
930948
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_exp(<8 x double> [[TMP4:%.*]])
949+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.exp.v16f64(<16 x double> [[TMP4:%.*]])
931950
; CHECK: ret void
932951
;
933952
entry:
@@ -950,6 +969,7 @@ for.end:
950969

951970
define void @exp_f32_intrinsic(ptr nocapture %varray) {
952971
; CHECK-LABEL: @exp_f32_intrinsic(
972+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.exp.v2f32(<2 x float> [[TMP4:%.*]])
953973
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_expf(<4 x float> [[TMP4:%.*]])
954974
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_expf(<8 x float> [[TMP4:%.*]])
955975
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_expf(<16 x float> [[TMP4:%.*]])
@@ -978,6 +998,7 @@ define void @log_f64(ptr nocapture %varray) {
978998
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_log(<2 x double> [[TMP4:%.*]])
979999
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_log(<4 x double> [[TMP4:%.*]])
9801000
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_log(<8 x double> [[TMP4:%.*]])
1001+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.log.v16f64(<16 x double> [[TMP4:%.*]])
9811002
; CHECK: ret void
9821003
;
9831004
entry:
@@ -1000,6 +1021,7 @@ for.end:
10001021

10011022
define void @log_f32(ptr nocapture %varray) {
10021023
; CHECK-LABEL: @log_f32(
1024+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.log.v2f32(<2 x float> [[TMP4:%.*]])
10031025
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_logf(<4 x float> [[TMP4:%.*]])
10041026
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_logf(<8 x float> [[TMP4:%.*]])
10051027
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_logf(<16 x float> [[TMP4:%.*]])
@@ -1028,6 +1050,7 @@ define void @log_f64_intrinsic(ptr nocapture %varray) {
10281050
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_log(<2 x double> [[TMP4:%.*]])
10291051
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_log(<4 x double> [[TMP4:%.*]])
10301052
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_log(<8 x double> [[TMP4:%.*]])
1053+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.log.v16f64(<16 x double> [[TMP4:%.*]])
10311054
; CHECK: ret void
10321055
;
10331056
entry:
@@ -1050,6 +1073,7 @@ for.end:
10501073

10511074
define void @log_f32_intrinsic(ptr nocapture %varray) {
10521075
; CHECK-LABEL: @log_f32_intrinsic(
1076+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.log.v2f32(<2 x float> [[TMP4:%.*]])
10531077
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_logf(<4 x float> [[TMP4:%.*]])
10541078
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_logf(<8 x float> [[TMP4:%.*]])
10551079
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_logf(<16 x float> [[TMP4:%.*]])
@@ -1078,6 +1102,7 @@ define void @log2_f64(ptr nocapture %varray) {
10781102
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_log2(<2 x double> [[TMP4:%.*]])
10791103
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_log2(<4 x double> [[TMP4:%.*]])
10801104
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_log2(<8 x double> [[TMP4:%.*]])
1105+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.log2.v16f64(<16 x double> [[TMP4:%.*]])
10811106
; CHECK: ret void
10821107
;
10831108
entry:
@@ -1100,6 +1125,7 @@ for.end:
11001125

11011126
define void @log2_f32(ptr nocapture %varray) {
11021127
; CHECK-LABEL: @log2_f32(
1128+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.log2.v2f32(<2 x float> [[TMP4:%.*]])
11031129
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_log2f(<4 x float> [[TMP4:%.*]])
11041130
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_log2f(<8 x float> [[TMP4:%.*]])
11051131
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_log2f(<16 x float> [[TMP4:%.*]])
@@ -1128,6 +1154,7 @@ define void @log2_f64_intrinsic(ptr nocapture %varray) {
11281154
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_log2(<2 x double> [[TMP4:%.*]])
11291155
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_log2(<4 x double> [[TMP4:%.*]])
11301156
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_log2(<8 x double> [[TMP4:%.*]])
1157+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.log2.v16f64(<16 x double> [[TMP4:%.*]])
11311158
; CHECK: ret void
11321159
;
11331160
entry:
@@ -1150,6 +1177,7 @@ for.end:
11501177

11511178
define void @log2_f32_intrinsic(ptr nocapture %varray) {
11521179
; CHECK-LABEL: @log2_f32_intrinsic(
1180+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.log2.v2f32(<2 x float> [[TMP4:%.*]])
11531181
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_log2f(<4 x float> [[TMP4:%.*]])
11541182
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_log2f(<8 x float> [[TMP4:%.*]])
11551183
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_log2f(<16 x float> [[TMP4:%.*]])
@@ -1175,6 +1203,7 @@ for.end:
11751203

11761204
define void @log10_f32(ptr nocapture %varray) {
11771205
; CHECK-LABEL: @log10_f32(
1206+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.log10.v2f32(<2 x float> [[TMP4:%.*]])
11781207
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_log10f(<4 x float> [[TMP4:%.*]])
11791208
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_log10f(<8 x float> [[TMP4:%.*]])
11801209
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_log10f(<16 x float> [[TMP4:%.*]])
@@ -1200,6 +1229,7 @@ for.end:
12001229

12011230
define void @log10_f32_intrinsic(ptr nocapture %varray) {
12021231
; CHECK-LABEL: @log10_f32_intrinsic(
1232+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.log10.v2f32(<2 x float> [[TMP4:%.*]])
12031233
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_log10f(<4 x float> [[TMP4:%.*]])
12041234
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_log10f(<8 x float> [[TMP4:%.*]])
12051235
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_log10f(<16 x float> [[TMP4:%.*]])
@@ -1228,6 +1258,7 @@ define void @exp2_f64(ptr nocapture %varray) {
12281258
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_exp2(<2 x double> [[TMP4:%.*]])
12291259
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_exp2(<4 x double> [[TMP4:%.*]])
12301260
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_exp2(<8 x double> [[TMP4:%.*]])
1261+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.exp2.v16f64(<16 x double> [[TMP4:%.*]])
12311262
; CHECK: ret void
12321263
;
12331264
entry:
@@ -1250,6 +1281,7 @@ for.end:
12501281

12511282
define void @exp2_f32(ptr nocapture %varray) {
12521283
; CHECK-LABEL: @exp2_f32(
1284+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.exp2.v2f32(<2 x float> [[TMP4:%.*]])
12531285
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_exp2f(<4 x float> [[TMP4:%.*]])
12541286
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_exp2f(<8 x float> [[TMP4:%.*]])
12551287
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_exp2f(<16 x float> [[TMP4:%.*]])
@@ -1278,6 +1310,7 @@ define void @exp2_f64_intrinsic(ptr nocapture %varray) {
12781310
; CHECK-VF2: [[TMP5:%.*]] = call <2 x double> @amd_vrd2_exp2(<2 x double> [[TMP4:%.*]])
12791311
; CHECK-VF4: [[TMP5:%.*]] = call <4 x double> @amd_vrd4_exp2(<4 x double> [[TMP4:%.*]])
12801312
; CHECK-VF8: [[TMP5:%.*]] = call <8 x double> @amd_vrd8_exp2(<8 x double> [[TMP4:%.*]])
1313+
; CHECK-VF16: [[TMP5:%.*]] = call <16 x double> @llvm.exp2.v16f64(<16 x double> [[TMP4:%.*]])
12811314
; CHECK: ret void
12821315
;
12831316
entry:
@@ -1300,6 +1333,7 @@ for.end:
13001333

13011334
define void @exp2_f32_intrinsic(ptr nocapture %varray) {
13021335
; CHECK-LABEL: @exp2_f32_intrinsic(
1336+
; CHECK-VF2: [[TMP5:%.*]] = call <2 x float> @llvm.exp2.v2f32(<2 x float> [[TMP4:%.*]])
13031337
; CHECK-VF4: [[TMP5:%.*]] = call <4 x float> @amd_vrs4_exp2f(<4 x float> [[TMP4:%.*]])
13041338
; CHECK-VF8: [[TMP5:%.*]] = call <8 x float> @amd_vrs8_exp2f(<8 x float> [[TMP4:%.*]])
13051339
; CHECK-VF16: [[TMP5:%.*]] = call <16 x float> @amd_vrs16_exp2f(<16 x float> [[TMP4:%.*]])

0 commit comments

Comments
 (0)