Skip to content

Commit c091792

Browse files
ecnelisesecnelises
committed
[NFC] Fix SSE3 intrinsics test for PowerPC
Previous test in ppc-pmmintrin.c did not check IR of intrinsic function definition. Add them and simplify. These tests shouldn't be auto-generated, because we don't want to check wrapper functions.
1 parent cd81723 commit c091792

File tree

1 file changed

+36
-45
lines changed

1 file changed

+36
-45
lines changed

‎clang/test/CodeGen/PowerPC/ppc-pmmintrin.c

Lines changed: 36 additions & 45 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,3 @@
1-
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
21
// REQUIRES: powerpc-registered-target
32

43
// RUN: %clang -S -emit-llvm -target powerpc64-gnu-linux -mcpu=pwr8 -DNO_MM_MALLOC -ffreestanding -DNO_WARN_X86_INTRINSICS %s \
@@ -19,53 +18,10 @@ __m128i resi, mi;
1918
double *d;
2019

2120
void __attribute__((noinline))
22-
// CHECK-LABEL: @test_pmmintrin(
23-
// CHECK-NEXT: entry:
24-
// CHECK-NEXT: [[TMP0:%.*]] = load <2 x double>, <2 x double>* @md1, align 16
25-
// CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, <2 x double>* @md2, align 16
26-
// CHECK-NEXT: [[CALL:%.*]] = call <2 x double> @_mm_addsub_pd(<2 x double> noundef [[TMP0]], <2 x double> noundef [[TMP1]]) #[[ATTR4:[0-9]+]]
27-
// CHECK-NEXT: store <2 x double> [[CALL]], <2 x double>* @resd, align 16
28-
// CHECK-NEXT: [[TMP2:%.*]] = load <4 x float>, <4 x float>* @m1, align 16
29-
// CHECK-NEXT: [[TMP3:%.*]] = load <4 x float>, <4 x float>* @m2, align 16
30-
// CHECK-NEXT: [[CALL1:%.*]] = call <4 x float> @_mm_addsub_ps(<4 x float> noundef [[TMP2]], <4 x float> noundef [[TMP3]]) #[[ATTR4]]
31-
// CHECK-NEXT: store <4 x float> [[CALL1]], <4 x float>* @res, align 16
32-
// CHECK-NEXT: [[TMP4:%.*]] = load <2 x double>, <2 x double>* @md1, align 16
33-
// CHECK-NEXT: [[TMP5:%.*]] = load <2 x double>, <2 x double>* @md2, align 16
34-
// CHECK-NEXT: [[CALL2:%.*]] = call <2 x double> @_mm_hadd_pd(<2 x double> noundef [[TMP4]], <2 x double> noundef [[TMP5]]) #[[ATTR4]]
35-
// CHECK-NEXT: store <2 x double> [[CALL2]], <2 x double>* @resd, align 16
36-
// CHECK-NEXT: [[TMP6:%.*]] = load <4 x float>, <4 x float>* @m1, align 16
37-
// CHECK-NEXT: [[TMP7:%.*]] = load <4 x float>, <4 x float>* @m2, align 16
38-
// CHECK-NEXT: [[CALL3:%.*]] = call <4 x float> @_mm_hadd_ps(<4 x float> noundef [[TMP6]], <4 x float> noundef [[TMP7]]) #[[ATTR4]]
39-
// CHECK-NEXT: store <4 x float> [[CALL3]], <4 x float>* @res, align 16
40-
// CHECK-NEXT: [[TMP8:%.*]] = load <2 x double>, <2 x double>* @md1, align 16
41-
// CHECK-NEXT: [[TMP9:%.*]] = load <2 x double>, <2 x double>* @md2, align 16
42-
// CHECK-NEXT: [[CALL4:%.*]] = call <2 x double> @_mm_hsub_pd(<2 x double> noundef [[TMP8]], <2 x double> noundef [[TMP9]]) #[[ATTR4]]
43-
// CHECK-NEXT: store <2 x double> [[CALL4]], <2 x double>* @resd, align 16
44-
// CHECK-NEXT: [[TMP10:%.*]] = load <4 x float>, <4 x float>* @m1, align 16
45-
// CHECK-NEXT: [[TMP11:%.*]] = load <4 x float>, <4 x float>* @m2, align 16
46-
// CHECK-NEXT: [[CALL5:%.*]] = call <4 x float> @_mm_hsub_ps(<4 x float> noundef [[TMP10]], <4 x float> noundef [[TMP11]]) #[[ATTR4]]
47-
// CHECK-NEXT: store <4 x float> [[CALL5]], <4 x float>* @res, align 16
48-
// CHECK-NEXT: [[CALL6:%.*]] = call <2 x i64> @_mm_lddqu_si128(<2 x i64>* noundef @mi) #[[ATTR4]]
49-
// CHECK-NEXT: store <2 x i64> [[CALL6]], <2 x i64>* @resi, align 16
50-
// CHECK-NEXT: [[TMP12:%.*]] = load double*, double** @d, align 8
51-
// CHECK-NEXT: [[CALL7:%.*]] = call <2 x double> @_mm_loaddup_pd(double* noundef [[TMP12]]) #[[ATTR4]]
52-
// CHECK-NEXT: store <2 x double> [[CALL7]], <2 x double>* @resd, align 16
53-
// CHECK-NEXT: [[TMP13:%.*]] = load <2 x double>, <2 x double>* @md1, align 16
54-
// CHECK-NEXT: [[CALL8:%.*]] = call <2 x double> @_mm_movedup_pd(<2 x double> noundef [[TMP13]]) #[[ATTR4]]
55-
// CHECK-NEXT: store <2 x double> [[CALL8]], <2 x double>* @resd, align 16
56-
// CHECK-NEXT: [[TMP14:%.*]] = load <4 x float>, <4 x float>* @m1, align 16
57-
// CHECK-NEXT: [[CALL9:%.*]] = call <4 x float> @_mm_movehdup_ps(<4 x float> noundef [[TMP14]]) #[[ATTR4]]
58-
// CHECK-NEXT: store <4 x float> [[CALL9]], <4 x float>* @res, align 16
59-
// CHECK-NEXT: [[TMP15:%.*]] = load <4 x float>, <4 x float>* @m1, align 16
60-
// CHECK-NEXT: [[CALL10:%.*]] = call <4 x float> @_mm_moveldup_ps(<4 x float> noundef [[TMP15]]) #[[ATTR4]]
61-
// CHECK-NEXT: store <4 x float> [[CALL10]], <4 x float>* @res, align 16
62-
// CHECK-NEXT: ret void
63-
//
6421
test_pmmintrin() {
6522
resd = _mm_addsub_pd(md1, md2);
6623
res = _mm_addsub_ps(m1, m2);
6724
resd = _mm_hadd_pd(md1, md2);
68-
//
6925
res = _mm_hadd_ps(m1, m2);
7026
resd = _mm_hsub_pd(md1, md2);
7127
res = _mm_hsub_ps(m1, m2);
@@ -76,14 +32,49 @@ test_pmmintrin() {
7632
res = _mm_moveldup_ps(m1);
7733
}
7834

35+
// CHECK-LABEL: define available_externally <2 x double> @_mm_addsub_pd(<2 x double> noundef %{{[0-9a-zA-Z_.]+}}, <2 x double> noundef %{{[0-9a-zA-Z_.]+}})
36+
// CHECK: call <2 x double> @vec_xor(double vector[2], double vector[2])(<2 x double> noundef %{{[0-9a-zA-Z_.]+}}, <2 x double> noundef <double -0.000000e+00, double 0.000000e+00>)
37+
// CHECK: call <2 x double> @vec_add(double vector[2], double vector[2])
7938

39+
// CHECK-LABEL: define available_externally <4 x float> @_mm_addsub_ps(<4 x float> noundef %{{[0-9a-zA-Z_.]+}}, <4 x float> noundef %{{[0-9a-zA-Z_.]+}})
40+
// CHECK: call <4 x float> @vec_xor(float vector[4], float vector[4])(<4 x float> noundef %{{[0-9a-zA-Z_.]+}}, <4 x float> noundef <float -0.000000e+00, float 0.000000e+00, float -0.000000e+00, float 0.000000e+00>)
41+
// CHECK: call <4 x float> @vec_add(float vector[4], float vector[4])
8042

43+
// CHECK-LABEL: define available_externally <2 x double> @_mm_hadd_pd(<2 x double> noundef %{{[0-9a-zA-Z_.]+}}, <2 x double> noundef %{{[0-9a-zA-Z_.]+}})
44+
// CHECK: %[[CALL:[0-9a-zA-Z_.]+]] = call <2 x double> @vec_mergeh(double vector[2], double vector[2])
45+
// CHECK: %[[CALL1:[0-9a-zA-Z_.]+]] = call <2 x double> @vec_mergel(double vector[2], double vector[2])
46+
// CHECK: %[[CALL2:[0-9a-zA-Z_.]+]] = call <2 x double> @vec_add(double vector[2], double vector[2])(<2 x double> noundef %[[CALL]], <2 x double> noundef %[[CALL1]])
8147

48+
// CHECK-LABEL: define available_externally <4 x float> @_mm_hadd_ps(<4 x float> noundef %{{[0-9a-zA-Z_.]+}}, <4 x float> noundef %{{[0-9a-zA-Z_.]+}})
49+
// CHECK: store <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 8, i8 9, i8 10, i8 11, i8 16, i8 17, i8 18, i8 19, i8 24, i8 25, i8 26, i8 27>, <16 x i8>* %{{[0-9a-zA-Z_.]+}}, align 16
50+
// CHECK: store <16 x i8> <i8 4, i8 5, i8 6, i8 7, i8 12, i8 13, i8 14, i8 15, i8 20, i8 21, i8 22, i8 23, i8 28, i8 29, i8 30, i8 31>, <16 x i8>* %{{[0-9a-zA-Z_.]+}}, align 16
51+
// CHECK: %[[CALL:[0-9a-zA-Z_.]+]] = call <4 x float> @vec_perm(float vector[4], float vector[4], unsigned char vector[16])
52+
// CHECK: %[[CALL1:[0-9a-zA-Z_.]+]] = call <4 x float> @vec_perm(float vector[4], float vector[4], unsigned char vector[16])
53+
// CHECK: %[[CALL2:[0-9a-zA-Z_.]+]] = call <4 x float> @vec_add(float vector[4], float vector[4])(<4 x float> noundef %[[CALL]], <4 x float> noundef %[[CALL1]])
8254

55+
// CHECK-LABEL: define available_externally <2 x double> @_mm_hsub_pd(<2 x double> noundef %{{[0-9a-zA-Z_.]+}}, <2 x double> noundef %{{[0-9a-zA-Z_.]+}})
56+
// CHECK: %[[CALL:[0-9a-zA-Z_.]+]] = call <2 x double> @vec_mergeh(double vector[2], double vector[2])
57+
// CHECK: %[[CALL1:[0-9a-zA-Z_.]+]] = call <2 x double> @vec_mergel(double vector[2], double vector[2])
58+
// CHECK: %[[CALL2:[0-9a-zA-Z_.]+]] = call <2 x double> @vec_sub(double vector[2], double vector[2])(<2 x double> noundef %[[CALL]], <2 x double> noundef %[[CALL1]])
8359

60+
// CHECK-LABEL: define available_externally <4 x float> @_mm_hsub_ps(<4 x float> noundef %{{[0-9a-zA-Z_.]+}}, <4 x float> noundef %{{[0-9a-zA-Z_.]+}})
61+
// CHECK: store <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 8, i8 9, i8 10, i8 11, i8 16, i8 17, i8 18, i8 19, i8 24, i8 25, i8 26, i8 27>, <16 x i8>* %{{[0-9a-zA-Z_.]+}}, align 16
62+
// CHECK: store <16 x i8> <i8 4, i8 5, i8 6, i8 7, i8 12, i8 13, i8 14, i8 15, i8 20, i8 21, i8 22, i8 23, i8 28, i8 29, i8 30, i8 31>, <16 x i8>* %{{[0-9a-zA-Z_.]+}}, align 16
63+
// CHECK: %[[CALL:[0-9a-zA-Z_.]+]] = call <4 x float> @vec_perm(float vector[4], float vector[4], unsigned char vector[16])
64+
// CHECK: %[[CALL1:[0-9a-zA-Z_.]+]] = call <4 x float> @vec_perm(float vector[4], float vector[4], unsigned char vector[16])
65+
// CHECK: %[[CALL2:[0-9a-zA-Z_.]+]] = call <4 x float> @vec_sub(float vector[4], float vector[4])(<4 x float> noundef %[[CALL]], <4 x float> noundef %[[CALL1]])
8466

67+
// CHECK-LABEL: define available_externally <2 x i64> @_mm_lddqu_si128(<2 x i64>* noundef %{{[0-9a-zA-Z_.]+}})
68+
// CHECK: call <4 x i32> @vec_vsx_ld(int, int const*)(i32 noundef signext 0, i32* noundef %{{[0-9a-zA-Z_.]+}})
8569

70+
// CHECK-LABEL: define available_externally <2 x double> @_mm_loaddup_pd(double* noundef %{{[0-9a-zA-Z_.]+}})
71+
// CHECK: call <2 x double> @vec_splats(double)
8672

73+
// CHECK-LABEL: define available_externally <2 x double> @_mm_movedup_pd(<2 x double> noundef %{{[0-9a-zA-Z_.]+}})
74+
// CHECK: call <2 x double> @_mm_shuffle_pd(<2 x double> noundef %{{[0-9a-zA-Z_.]+}}, <2 x double> noundef %{{[0-9a-zA-Z_.]+}}, i32 noundef signext 0)
8775

76+
// CHECK-LABEL: define available_externally <4 x float> @_mm_movehdup_ps(<4 x float> noundef %{{[0-9a-zA-Z_.]+}})
77+
// CHECK: call <4 x i32> @vec_mergeo(unsigned int vector[4], unsigned int vector[4])
8878

89-
79+
// CHECK-LABEL: define available_externally <4 x float> @_mm_moveldup_ps(<4 x float> noundef %{{[0-9a-zA-Z_.]+}})
80+
// CHECK: call <4 x i32> @vec_mergee(unsigned int vector[4], unsigned int vector[4])

0 commit comments

Comments
 (0)