Skip to content
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

Jump to bottom

Add framework for implementation of support for AVX512VNNI #3894

Draft
wants to merge 4 commits into
base: main
Choose a base branch
from
Draft

Add framework for implementation of support for AVX512VNNI #3894

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
99cc865
Test whether OS supports AVX512
stweil Aug 6, 2022
5703419
Implement intSimdMatrixAVX512VNNI (dummy)
stweil Aug 5, 2022
8d0bd68
Add unittest for AVX512VNNI
stweil Aug 6, 2022
232093f
Add the needed changes to support AVX512VNNI
amitdo Aug 9, 2022
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
8 changes: 8 additions & 0 deletions Makefile.am
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -170,6 +170,14 @@ libtesseract_la_LIBADD += libtesseract_avx512.la
noinst_LTLIBRARIES += libtesseract_avx512.la
endif

if HAVE_AVX512VNNI
libtesseract_avx512vnni_la_CXXFLAGS = -mavx512vnni -mavx512vl
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am curious: what is the use of -mavx512vl? Should it be checked in configure, too?

Copy link
Collaborator

@amitdo amitdo Aug 9, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It does not compile without the second flag. It is needed because the code uses ymm registers (256 bit) instead of zmm registers (512 bit).

Yes, we can also add another check in configure.

Note that AFAIK all Intel's CPUs that support AVX512VNNI also support AVXVL.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for explaining. Maybe we don't need the additional check then.

libtesseract_avx512vnni_la_CXXFLAGS += -I$(top_srcdir)/src/ccutil
libtesseract_avx512vnni_la_SOURCES = src/arch/intsimdmatrixavx512vnni.cpp
libtesseract_la_LIBADD += libtesseract_avx512vnni.la
noinst_LTLIBRARIES += libtesseract_avx512vnni.la
endif

if HAVE_FMA
libtesseract_fma_la_CXXFLAGS = -mfma
libtesseract_fma_la_CXXFLAGS += -I$(top_srcdir)/src/ccutil
Expand Down
7 changes: 7 additions & 0 deletions configure.ac
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,6 +130,7 @@ AX_CHECK_COMPILE_FLAG([-Werror=unused-command-line-argument], [WERROR=-Werror=un
AM_CONDITIONAL([HAVE_AVX], false)
AM_CONDITIONAL([HAVE_AVX2], false)
AM_CONDITIONAL([HAVE_AVX512F], false)
AM_CONDITIONAL([HAVE_AVX512VNNI], false)
AM_CONDITIONAL([HAVE_FMA], false)
AM_CONDITIONAL([HAVE_SSE4_1], false)
AM_CONDITIONAL([HAVE_NEON], false)
Expand All @@ -156,6 +157,12 @@ case "${host_cpu}" in
AC_DEFINE([HAVE_AVX512F], [1], [Enable AVX512F instructions])
fi

AX_CHECK_COMPILE_FLAG([-mavx512vnni], [avx512vnni=true], [avx512vnni=false], [$WERROR])
AM_CONDITIONAL([HAVE_AVX512VNNI], $avx512vnni)
if $avx512vnni; then
AC_DEFINE([HAVE_AVX512VNNI], [1], [Enable AVX512VNNI instructions])
fi

AX_CHECK_COMPILE_FLAG([-mfma], [fma=true], [fma=false], [$WERROR])
AM_CONDITIONAL([HAVE_FMA], $fma)
if $fma; then
Expand Down
1 change: 1 addition & 0 deletions src/arch/intsimdmatrix.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -117,6 +117,7 @@ struct TESS_API IntSimdMatrix {
static const IntSimdMatrix intSimdMatrixNEON;
// Only available with AVX2 / AVX / FMA / SSE.
static const IntSimdMatrix intSimdMatrixAVX2;
static const IntSimdMatrix intSimdMatrixAVX512VNNI;
static const IntSimdMatrix intSimdMatrixSSE;
};

Expand Down
Loading