blob: 1b4406380dec92f6c67015e5ccfdfd94bda363f7 (
plain) (
blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
|
ifeq ($(subdir),mathvec)
libmvec-support += svml_d_cos2_core svml_d_cos4_core_avx \
svml_d_cos4_core svml_d_cos8_core \
svml_d_sin2_core svml_d_sin4_core_avx \
svml_d_sin4_core svml_d_sin8_core svml_d_trig_data \
svml_s_cosf4_core svml_s_cosf8_core_avx \
svml_s_cosf8_core svml_s_cosf16_core svml_s_trig_data \
svml_s_sinf4_core svml_s_sinf8_core_avx \
svml_s_sinf8_core svml_s_sinf16_core \
svml_d_sincos2_core svml_d_sincos4_core_avx \
svml_d_sincos4_core svml_d_sincos8_core \
svml_d_log2_core svml_d_log4_core_avx svml_d_log4_core \
svml_d_log8_core svml_d_log_data svml_s_logf4_core \
svml_s_logf8_core_avx svml_s_logf8_core svml_s_logf16_core \
svml_s_logf_data svml_d_exp2_core svml_d_exp4_core_avx \
svml_d_exp4_core svml_d_exp8_core svml_d_exp_data \
svml_s_expf4_core svml_s_expf8_core_avx svml_s_expf8_core \
svml_s_expf16_core svml_s_expf_data svml_d_pow2_core \
svml_d_pow4_core_avx svml_d_pow4_core svml_d_pow8_core \
svml_d_pow_data svml_s_powf4_core svml_s_powf8_core_avx \
svml_s_powf8_core svml_s_powf16_core svml_s_powf_data \
svml_s_sincosf4_core svml_s_sincosf8_core_avx \
svml_s_sincosf8_core svml_s_sincosf16_core
endif
# Variables for libmvec tests.
ifeq ($(subdir),math)
ifeq ($(build-mathvec),yes)
libmvec-tests += double-vlen2 double-vlen4 double-vlen4-avx2 \
float-vlen4 float-vlen8 float-vlen8-avx2
tests += test-double-libmvec-sincos test-double-libmvec-sincos-avx \
test-double-libmvec-sincos-avx2 test-float-libmvec-sincosf \
test-float-libmvec-sincosf-avx test-float-libmvec-sincosf-avx2
test-extras += test-double-libmvec-sincos-avx-main \
test-double-libmvec-sincos-avx2-main \
test-double-libmvec-sincos-main \
test-float-libmvec-sincosf-avx-main \
test-float-libmvec-sincosf-avx2-main \
test-float-libmvec-sincosf-main
extra-test-objs += test-double-libmvec-sincos-avx-main.o \
test-double-libmvec-sincos-avx2-main.o \
test-double-libmvec-sincos-main.o \
test-float-libmvec-sincosf-avx-main.o \
test-float-libmvec-sincosf-avx2-main.o\
test-float-libmvec-sincosf-main.o
$(objpfx)test-double-libmvec-sincos: \
$(objpfx)test-double-libmvec-sincos.o \
$(objpfx)test-double-libmvec-sincos-main.o $(libmvec)
$(objpfx)test-double-libmvec-sincos-avx: \
$(objpfx)test-double-libmvec-sincos-avx.o \
$(objpfx)test-double-libmvec-sincos-avx-main.o $(libmvec)
$(objpfx)test-double-libmvec-sincos-avx2: \
$(objpfx)test-double-libmvec-sincos-avx2.o \
$(objpfx)test-double-libmvec-sincos-avx2-main.o $(libmvec)
$(objpfx)test-float-libmvec-sincosf: \
$(objpfx)test-float-libmvec-sincosf.o \
$(objpfx)test-float-libmvec-sincosf-main.o $(libmvec)
$(objpfx)test-float-libmvec-sincosf-avx: \
$(objpfx)test-float-libmvec-sincosf-avx.o \
$(objpfx)test-float-libmvec-sincosf-avx-main.o $(libmvec)
$(objpfx)test-float-libmvec-sincosf-avx2: \
$(objpfx)test-float-libmvec-sincosf-avx2.o \
$(objpfx)test-float-libmvec-sincosf-avx2-main.o $(libmvec)
libmvec-tests += double-vlen8 float-vlen16
tests += test-double-libmvec-sincos-avx512 \
test-float-libmvec-sincosf-avx512
test-extras += test-double-libmvec-sincos-avx512-main \
test-float-libmvec-sincosf-avx512-main
extra-test-objs += test-double-libmvec-sincos-avx512-main.o \
test-float-libmvec-sincosf-avx512-main.o
$(objpfx)test-double-libmvec-sincos-avx512: \
$(objpfx)test-double-libmvec-sincos-avx512.o \
$(objpfx)test-double-libmvec-sincos-avx512-main.o $(libmvec)
$(objpfx)test-float-libmvec-sincosf-avx512: \
$(objpfx)test-float-libmvec-sincosf-avx512.o \
$(objpfx)test-float-libmvec-sincosf-avx512-main.o $(libmvec)
double-vlen2-funcs = cos exp log pow sin sincos
double-vlen4-funcs = cos exp log pow sin sincos
double-vlen4-avx2-funcs = cos exp log pow sin sincos
double-vlen8-funcs = cos exp log pow sin sincos
float-vlen4-funcs = cos exp log pow sin sincos
float-vlen8-funcs = cos exp log pow sin sincos
float-vlen8-avx2-funcs = cos exp log pow sin sincos
float-vlen16-funcs = cos exp log pow sin sincos
double-vlen4-arch-ext-cflags = -mavx
double-vlen4-arch-ext2-cflags = -mavx2
double-vlen8-arch-ext-cflags = -mavx512f
float-vlen8-arch-ext-cflags = -mavx
float-vlen8-arch-ext2-cflags = -mavx2
float-vlen16-arch-ext-cflags = -mavx512f
libmvec-sincos-cflags = $(libm-test-fast-math-cflags) -fno-inline -fopenmp -Wno-unknown-pragmas
CFLAGS-test-double-vlen4-avx2-wrappers.c = $(double-vlen4-arch-ext2-cflags)
CFLAGS-test-float-vlen8-avx2-wrappers.c = $(float-vlen8-arch-ext2-cflags)
CFLAGS-test-double-libmvec-sincos-main.c = $(libmvec-sincos-cflags)
CFLAGS-test-double-libmvec-sincos-avx.c = -DREQUIRE_AVX
CFLAGS-test-double-libmvec-sincos-avx-main.c = $(libmvec-sincos-cflags) $(double-vlen4-arch-ext-cflags)
CFLAGS-test-double-libmvec-sincos-avx2.c = -DREQUIRE_AVX2
CFLAGS-test-double-libmvec-sincos-avx2-main.c = $(libmvec-sincos-cflags) $(double-vlen4-arch-ext2-cflags)
CFLAGS-test-double-libmvec-sincos-avx512.c = -DREQUIRE_AVX512F
CFLAGS-test-double-libmvec-sincos-avx512-main.c = $(libmvec-sincos-cflags) $(double-vlen8-arch-ext-cflags)
CFLAGS-test-float-libmvec-sincosf-main.c = $(libmvec-sincos-cflags)
CFLAGS-test-float-libmvec-sincosf-avx.c = -DREQUIRE_AVX
CFLAGS-test-float-libmvec-sincosf-avx-main.c = $(libmvec-sincos-cflags) $(float-vlen8-arch-ext-cflags)
CFLAGS-test-float-libmvec-sincosf-avx2.c = -DREQUIRE_AVX2
CFLAGS-test-float-libmvec-sincosf-avx2-main.c = $(libmvec-sincos-cflags) $(float-vlen8-arch-ext2-cflags)
CFLAGS-test-float-libmvec-sincosf-avx512.c = -DREQUIRE_AVX512F
CFLAGS-test-float-libmvec-sincosf-avx512-main.c = $(libmvec-sincos-cflags) $(float-vlen16-arch-ext-cflags)
endif
endif
ifeq ($(subdir)$(config-cflags-mprefer-vector-width),mathyes)
# When compiled with -O3 -march=skylake, GCC 8 and 9 optimize some loops
# in branred.c with 256-bit vector instructions, which leads to store
# forward stall:
#
# https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90579
#
# Limit vector width to 128 bits to work around this issue. It improves
# performance of sin and cos by more than 40% on Skylake.
CFLAGS-branred.c = -mprefer-vector-width=128
endif
|
