blob: f683b62b8f4c06893174b6e84236c5fdec00a6e9 (
plain)
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
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
|
# Copyright (c) 2018 Cisco and/or its affiliates.
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at:
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
macro(set_log2_cacheline_size var n)
if(${n} EQUAL 128)
set(${var} 7)
elseif(${n} EQUAL 64)
set(${var} 6)
else()
message(FATAL_ERROR "Cacheline size ${n} not supported")
endif()
endmacro()
##############################################################################
# Cache line size
##############################################################################
if(DEFINED VPP_CACHE_LINE_SIZE)
# Cache line size assigned via cmake args
elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(aarch64.*|AARCH64.*)")
set(VPP_CACHE_LINE_SIZE 128)
else()
set(VPP_CACHE_LINE_SIZE 64)
endif()
set(VPP_CACHE_LINE_SIZE ${VPP_CACHE_LINE_SIZE}
CACHE STRING "Target CPU cache line size")
set_log2_cacheline_size(VPP_LOG2_CACHE_LINE_SIZE ${VPP_CACHE_LINE_SIZE})
##############################################################################
# Gnu Assembler AVX-512 bug detection
# - see: https://sourceware.org/bugzilla/show_bug.cgi?id=23465
##############################################################################
if(CMAKE_SYSTEM_PROCESSOR MATCHES "amd64.*|x86_64.*|AMD64.*")
if (CMAKE_C_COMPILER_ID STREQUAL "GNU")
set(pfx ${CMAKE_CURRENT_BINARY_DIR}${CMAKE_FILES_DIRECTORY}/gas_avx512_bug_test)
file(WRITE ${pfx}.s "vmovaps 0x40(,%rax), %zmm0\n")
execute_process(COMMAND ${CMAKE_C_COMPILER} -c ${pfx}.s -o ${pfx}.o)
execute_process(COMMAND objdump -s ${pfx}.o OUTPUT_VARIABLE _output)
if (NOT _output MATCHES "62f17c48 28040540 000000")
set(GNU_ASSEMBLER_AVX512_BUG 1)
endif()
endif()
endif()
##############################################################################
# CPU optimizations and multiarch support
##############################################################################
option(VPP_BUILD_NATIVE_ONLY "Build only for native CPU." OFF)
macro(add_vpp_march_variant v)
cmake_parse_arguments(ARG
"OFF"
"N_PREFETCHES;CACHE_PREFETCH_BYTES"
"FLAGS"
${ARGN}
)
if(ARG_FLAGS)
set(flags_ok 1)
set(fs "")
foreach(f ${ARG_FLAGS})
string(APPEND fs " ${f}")
string(REGEX REPLACE "[-=+]" "_" sfx ${f})
if(NOT DEFINED compiler_flag${sfx})
check_c_compiler_flag(${f} compiler_flag${sfx})
endif()
if(NOT compiler_flag${sfx})
unset(flags_ok)
endif()
endforeach()
if(ARG_N_PREFETCHES)
string(APPEND fs " -DCLIB_N_PREFETCHES=${ARG_N_PREFETCHES}")
endif()
if(ARG_CACHE_PREFETCH_BYTES)
set_log2_cacheline_size(log2 ${ARG_CACHE_PREFETCH_BYTES})
string(APPEND fs " -DCLIB_LOG2_CACHE_PREFETCH_BYTES=${log2}")
endif()
if(flags_ok)
string(TOUPPER ${v} uv)
if(ARG_OFF)
option(VPP_MARCH_VARIANT_${uv} "Build ${v} multiarch variant." OFF)
else()
option(VPP_MARCH_VARIANT_${uv} "Build ${v} multiarch variant." ON)
endif()
if (VPP_MARCH_VARIANT_${uv})
list(APPEND MARCH_VARIANTS "${v}\;${fs}")
list(APPEND MARCH_VARIANTS_NAMES "${v}")
else()
list(APPEND MARCH_VARIANTS_DISABLED "${v}\;${fs}")
endif()
endif()
endif()
endmacro()
if(VPP_BUILD_NATIVE_ONLY)
set(VPP_BUILD_NATIVE_ARCH "native" CACHE STRING "native CPU -march= value.")
set(VPP_DEFAULT_MARCH_FLAGS -march=${VPP_BUILD_NATIVE_ARCH})
if(VPP_BUILD_NATIVE_ONLY)
check_c_compiler_flag(${VPP_DEFAULT_MARCH_FLAGS} compiler_flag_march)
if(NOT compiler_flag_march)
message(FATAL_ERROR "Native-only build with ${VPP_DEFAULT_MARCH_FLAGS} is not supported by compiler")
endif()
endif()
set(MARCH_VARIANTS_NAMES "native-only")
elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "amd64.*|x86_64.*|AMD64.*")
set(VPP_DEFAULT_MARCH_FLAGS -march=corei7 -mtune=corei7-avx)
add_vpp_march_variant(hsw
FLAGS -march=haswell -mtune=haswell
)
add_vpp_march_variant(trm
FLAGS -march=tremont -mtune=tremont
OFF
)
add_vpp_march_variant(adl
FLAGS -march=alderlake -mtune=alderlake -mprefer-vector-width=256
OFF
)
add_vpp_march_variant(scalar
FLAGS -march=core2 -mno-mmx -mno-sse
OFF
)
add_vpp_march_variant(znver3
FLAGS -march=znver3 -mtune=znver3 -mprefer-vector-width=256
OFF
)
if (GNU_ASSEMBLER_AVX512_BUG)
message(WARNING "AVX-512 multiarch variant(s) disabled due to GNU Assembler bug")
else()
add_vpp_march_variant(skx
FLAGS -march=skylake-avx512 -mtune=skylake-avx512 -mprefer-vector-width=256
)
add_vpp_march_variant(icl
FLAGS -march=icelake-client -mtune=icelake-client -mprefer-vector-width=512
)
add_vpp_march_variant(spr
FLAGS -march=sapphirerapids -mtune=sapphirerapids -mprefer-vector-width=512
OFF
)
add_vpp_march_variant(znver4
FLAGS -march=znver4 -mtune=znver4 -mprefer-vector-width=512
OFF
)
endif()
elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(aarch64.*|AARCH64.*)")
set(VPP_DEFAULT_MARCH_FLAGS -march=armv8-a+crc)
add_vpp_march_variant(qdf24xx
FLAGS -march=armv8-a+crc+crypto -mtune=qdf24xx
N_PREFETCHES 8
CACHE_PREFETCH_BYTES 64
OFF
)
add_vpp_march_variant(octeontx2
FLAGS -march=armv8.2-a+crc+crypto+lse
N_PREFETCHES 8
)
add_vpp_march_variant(thunderx2t99
FLAGS -march=armv8.1-a+crc+crypto -mtune=thunderx2t99
N_PREFETCHES 8
CACHE_PREFETCH_BYTES 64
)
add_vpp_march_variant(cortexa72
FLAGS -march=armv8-a+crc+crypto -mtune=cortex-a72
N_PREFETCHES 6
CACHE_PREFETCH_BYTES 64
)
add_vpp_march_variant(neoversen1
FLAGS -march=armv8.2-a+crc+crypto -mtune=neoverse-n1
N_PREFETCHES 6
CACHE_PREFETCH_BYTES 64
)
add_vpp_march_variant(neoversen2
FLAGS -march=armv8.6-a+crc+crypto+sve2 -mtune=neoverse-n2
N_PREFETCHES 8
CACHE_PREFETCH_BYTES 64
OFF
)
endif()
macro(vpp_library_set_multiarch_sources lib)
cmake_parse_arguments(ARG
""
""
"SOURCES;DEPENDS;FORCE_ON"
${ARGN}
)
set(VARIANTS "${MARCH_VARIANTS}")
if(ARG_FORCE_ON)
foreach(F ${ARG_FORCE_ON})
foreach(V ${MARCH_VARIANTS_DISABLED})
list(GET V 0 VARIANT)
if (VARIANT STREQUAL F)
list(GET V 1 VARIANT_FLAGS)
list(APPEND VARIANTS "${VARIANT}\;${VARIANT_FLAGS}")
endif()
endforeach()
endforeach()
endif()
foreach(V ${VARIANTS})
list(GET V 0 VARIANT)
list(GET V 1 VARIANT_FLAGS)
set(l ${lib}_${VARIANT})
add_library(${l} OBJECT ${ARG_SOURCES})
if(ARG_DEPENDS)
add_dependencies(${l} ${ARG_DEPENDS})
endif()
set_target_properties(${l} PROPERTIES POSITION_INDEPENDENT_CODE ON)
target_compile_definitions(${l} PUBLIC CLIB_MARCH_VARIANT=${VARIANT})
separate_arguments(VARIANT_FLAGS)
target_compile_options(${l} PUBLIC ${VARIANT_FLAGS})
target_sources(${lib} PRIVATE $<TARGET_OBJECTS:${l}>)
endforeach()
endmacro()
|
