8000 Add detection code for avx/avx2/etc by howard0su · Pull Request #809 · ggml-org/llama.cpp · GitHub
[go: up one dir, main page]
Skip to content

Add detection code for avx/avx2/etc #809

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 4 commits into from
Closed

Add detection code for avx/avx2/etc #809

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
5f50d15
Add detection code for avx
howard0su Apr 1, 2023
7adce4f
Only check hardware when option is ON
howard0su Apr 7, 2023
ac072d7
Modify per code review sugguestions
howard0su Apr 16, 2023
a8a22ff
Build locally will detect CPU
howard0su Jun 1, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Modify per code review sugguestions
  • Loading branch information
@howard0su
howard0su committed Jun 1, 2023
commit ac072d7c91ca6be205bbcfa0922418027d93f1a9
108 changes: 3 additions & 105 deletions CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -109,111 +109,9 @@ else()
message(WARNING "Git repository not found; to enable automatic generation of build info, make sure Git is installed and the project is a Git repository.")
endif()

INCLUDE(CheckCSourceRuns)

SET(AVX_CODE "
#include <immintrin.h>
int main()
{
__m256 a;
a = _mm256_set1_ps(0);
return 0;
}
")

SET(AVX512_CODE "
#include <immintrin.h>
int main()
{
__m512i a = _mm512_set_epi8(0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0);
__m512i b = a;
__mmask64 equality_mask = _mm512_cmp_epi8_mask(a, b, _MM_CMPINT_EQ);
return 0;
}
")

SET(AVX2_CODE "
#include <immintrin.h>
int main()
{
__m256i a = {0};
a = _mm256_abs_epi16(a);
__m256i x;
_mm256_extract_epi64(x, 0); // we rely on this in our AVX2 code
return 0;
}
")

SET(FMA_CODE "
#include <immintrin.h>
int main()
{
__m256 acc = _mm256_setzero_ps();
const __m256 d = _mm256_setzero_ps();
const __m256 p = _mm256_setzero_ps();
acc = _mm256_fmadd_ps( d, p, acc );
return 0;
}
")

MACRO(CHECK_SSE type flags)
SET(__FLAG_I 1)
SET(CMAKE_REQUIRED_FLAGS_SAVE ${CMAKE_REQUIRED_FLAGS})
FOREACH(__FLAG ${flags})
IF(NOT ${type}_FOUND)
SET(CMAKE_REQUIRED_FLAGS ${__FLAG})
CHECK_C_SOURCE_RUNS("${${type}_CODE}" HAS_${type}_${__FLAG_I})
IF(HAS_${type}_${__FLAG_I})
SET(${type}_FOUND TRUE CACHE BOOL "${type} support")
SET(${type}_FLAGS "${__FLAG}" CACHE STRING "${type} flags")
ENDIF()
MATH(EXPR __FLAG_I "${__FLAG_I}+1")
ENDIF()
ENDFOREACH()
SET(CMAKE_REQUIRED_FLAGS ${CMAKE_REQUIRED_FLAGS_SAVE})

IF(NOT ${type}_FOUND)
SET(${type}_FOUND FALSE CACHE BOOL "${type} support")
SET(${type}_FLAGS "" CACHE STRING "${type} flags")
ENDIF()

MARK_AS_ADVANCED(${type}_FOUND ${type}_FLAGS)

ENDMACRO()

IF(${LLAMA_AVX})
CHECK_SSE("AVX" " ;-mavx;/arch:AVX")
IF(NOT ${AVX_FOUND})
set(LLAMA_AVX OFF)
ENDIF()
ENDIF()

IF(${LLAMA_AVX2})
CHECK_SSE("AVX2" " ;-mavx2 -mfma;/arch:AVX2")
IF(NOT ${AVX2_FOUND})
set(LLAMA_AVX2 OFF)
ENDIF()
ENDIF()

IF(${LLAMA_AVX512})
CHECK_SSE("AVX512" " ;-mavx512f -mavx512dq -mavx512vl -mavx512bw -mfma;/arch:AVX512")
IF(NOT ${AVX512_FOUND})
set(LLAMA_AVX512 OFF)
ENDIF()
ENDIF()

IF(${LLAMA_FMA})
CHECK_SSE("FMA" " ;-mfma;")
IF (NOT ${FMA_FOUND})
set(LLAMA_FMA OFF)
ENDIF()
MESSAGE("NATIVE=" ${LLAMA_NATIVE} " MSVC=" ${MSVC})
IF(LLAMA_NATIVE AND MSVC)
include(cmake/FindSIMD.cmake)
ENDIF()

#
Expand Down
99 changes: 99 additions & 0 deletions cmake/FindSIMD.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,99 @@
INCLUDE(CheckCSourceRuns)

SET(AVX_CODE "
#include <immintrin.h>
int main()
{
__m256 a;
a = _mm256_set1_ps(0);
return 0;
}
")

SET(AVX512_CODE "
#include <immintrin.h>
int main()
{
__m512i a = _mm512_set_epi8(0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0);
__m512i b = a;
__mmask64 equality_mask = _mm512_cmp_epi8_mask(a, b, _MM_CMPINT_EQ);
return 0;
}
")

SET(AVX2_CODE "
#include <immintrin.h>
int main()
{
__m256i a = {0};
a = _mm256_abs_epi16(a);
__m256i x;
_mm256_extract_epi64(x, 0); // we rely on this in our AVX2 code
return 0;
}
")

SET(FMA_CODE "
#include <immintrin.h>
int main()
{
__m256 acc = _mm256_setzero_ps();
const __m256 d = _mm256_setzero_ps();
const __m256 p = _mm256_setzero_ps();
acc = _mm256_fmadd_ps( d, p, acc );
return 0;
}
")

MACRO(CHECK_SSE type flags)
SET(__FLAG_I 1)
SET(CMAKE_REQUIRED_FLAGS_SAVE ${CMAKE_REQUIRED_FLAGS})
FOREACH(__FLAG ${flags})
IF(NOT ${type}_FOUND)
SET(CMAKE_REQUIRED_FLAGS ${__FLAG})
CHECK_C_SOURCE_RUNS("${${type}_CODE}" HAS_${type}_${__FLAG_I})
IF(HAS_${type}_${__FLAG_I})
SET(${type}_FOUND TRUE CACHE BOOL "${type} support")
SET(${type}_FLAGS "${__FLAG}" CACHE STRING "${type} flags")
ENDIF()
MATH(EXPR __FLAG_I "${__FLAG_I}+1")
ENDIF()
ENDFOREACH()
SET(CMAKE_REQUIRED_FLAGS ${CMAKE_REQUIRED_FLAGS_SAVE})

IF(NOT ${type}_FOUND)
SET(${type}_FOUND FALSE CACHE BOOL "${type} support")
SET(${type}_FLAGS "" CACHE STRING "${type} flags")
ENDIF()

MARK_AS_ADVANCED(${type}_FOUND ${type}_FLAGS)

ENDMACRO()

CHECK_SSE("AVX" " ;/arch:AVX")
IF(NOT ${AVX_FOUND})
set(LLAMA_AVX OFF)
ELSE()
set(LLAMA_AVX ON)
ENDIF()

CHECK_SSE("AVX2" " ;/arch:AVX2")
IF(NOT ${AVX2_FOUND})
set(LLAMA_AVX2 OFF)
ELSE()
set(LLAMA_AVX2 ON)
ENDIF()

CHECK_SSE("AVX512" " ;/arch:AVX512")
IF(NOT ${AVX512_FOUND})
set(LLAMA_AVX512 OFF)
ELSE()
set(LLAMA_AVX512 ON)
ENDIF()
0