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
|
ifeq ($(subdir),mathvec)
libmvec-double-func-list = \
2_core \
4_core \
4_core_avx \
8_core
libmvec-float-func-list = \
f4_core \
f8_core \
f8_core_avx \
f16_core
libmvec-support += \
svml_d_exp_data \
svml_d_log_data \
svml_d_pow_data \
svml_d_trig_data \
svml_s_expf_data \
svml_s_logf_data \
svml_s_powf_data \
svml_s_trig_data \
$(foreach l,$(libmvec-double-func-list), \
$(addprefix svml_d_,$(addsuffix $(l),$(libmvec-funcs)))) \
$(foreach l,$(libmvec-float-func-list), \
$(addprefix svml_s_,$(addsuffix $(l),$(libmvec-funcs))))
endif
# Variables for libmvec tests.
ifeq ($(subdir)$(build-mathvec),mathyes)
libmvec-tests += double-vlen2 double-vlen4 double-vlen4-avx2 \
float-vlen4 float-vlen8 float-vlen8-avx2 \
double-vlen8 float-vlen16
tests += \
$(libmvec-abi-func-tests) \
$(libmvec-abi-func-avx-tests) \
$(libmvec-abi-func-avx2-tests) \
$(libmvec-abi-func-avx512f-tests)
double-vlen2-funcs = $(libmvec-funcs)
double-vlen4-funcs = $(libmvec-funcs)
double-vlen4-avx2-funcs = $(libmvec-funcs)
double-vlen8-funcs = $(libmvec-funcs)
float-vlen4-funcs = $(libmvec-funcs)
float-vlen8-funcs = $(libmvec-funcs)
float-vlen8-avx2-funcs = $(libmvec-funcs)
float-vlen16-funcs = $(libmvec-funcs)
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-abi-test-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 = $(float-vlen8-arch-ext2-cflags)
endif
ifeq ($(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
|