CI: Fix CI SIMD build on s390x #21750
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
upgrades to Jammy leads to different error: INFO: compile options: '-Inumpy/core/src/common -Inumpy/core/src -Inumpy/core -Inumpy/core/src/npymath -Inumpy/core/src/multiarray -Inumpy/core/src/umath -Inumpy/core/src/npysort -Inumpy/core/src/_simd -I/home/travis/build/numpy/numpy/builds/venv/include -I/opt/python/3.10.4/include/python3.10 -Ibuild/src.linux-s390x-3.10/numpy/core/src/common -Ibuild/src.linux-s390x-3.10/numpy/core/src/npymath -c'
extra options: '-march=arch11 -mzvector -Werror'
WARN: CCompilerOpt.dist_test[630] : CCompilerOpt._dist_test_spawn[764] : Command (gcc -pthread -Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Werror=vla -Werror=nonnull -Werror=pointer-arith -Werror=implicit-function-declaration -Wlogical-op -Wno-sign-compare -Werror=undef -fPIC -Inumpy/core/src/common -Inumpy/core/src -Inumpy/core -Inumpy/core/src/npymath -Inumpy/core/src/multiarray -Inumpy/core/src/umath -Inumpy/core/src/npysort -Inumpy/core/src/_simd -I/home/travis/build/numpy/numpy/builds/venv/include -I/opt/python/3.10.4/include/python3.10 -Ibuild/src.linux-s390x-3.10/numpy/core/src/common -Ibuild/src.linux-s390x-3.10/numpy/core/src/npymath -c /home/travis/build/numpy/numpy/numpy/distutils/checks/cpu_vx.c -o /tmp/tmp4dhgcm3y/home/travis/build/numpy/numpy/numpy/distutils/checks/cpu_vx.o -MMD -MF /tmp/tmp4dhgcm3y/home/travis/build/numpy/numpy/numpy/distutils/checks/cpu_vx.o.d -march=arch11 -mzvector -Werror) failed with exit status 1 output ->
/home/travis/build/numpy/numpy/numpy/distutils/checks/cpu_vx.c: In function ‘main’:
/home/travis/build/numpy/numpy/numpy/distutils/checks/cpu_vx.c:11:5: error: vector argument passed to unprototyped function
x = vec_round(vec_ceil(x) + vec_floor(y));
|
seiko2plus
force-pushed
the
fix_ci_ibmz_assembler
branch
7 times, most recently
from
ca8f9e1
to
df408ec
Compare
It was bionic, Travis doesn't provides jammy on s390x |
seiko2plus
force-pushed
the
fix_ci_ibmz_assembler
branch
4 times, most recently
from
5c85f5d
to
f800fca
Compare
|
upgrade the binutils fixes the issue and VX z13 build error can be triggered : build errorloops_comparison.dispatch.c.src: In function ‘simd_binary_equal_b8’:
numpy/core/src/common/simd/vec/operators.h:148:22: error: ‘__builtin_s390_vec_eqv’ requires z14 or higher
148 | #define npyv_xnor_b8 vec_eqv
| ^~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src:236:21: note: in expansion of macro ‘npyv_xnor_b8’
236 | npyv_b8 c = npyv_@VOP@_b8(a, b);
| ^~~~~~~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src: In function ‘simd_binary_scalar1_equal_b8’:
numpy/core/src/common/simd/vec/operators.h:148:22: error: ‘__builtin_s390_vec_eqv’ requires z14 or higher
148 | #define npyv_xnor_b8 vec_eqv
| ^~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src:260:21: note: in expansion of macro ‘npyv_xnor_b8’
260 | npyv_b8 c = npyv_@VOP@_b8(a, b);
| ^~~~~~~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src: In function ‘simd_binary_scalar2_equal_b8’:
numpy/core/src/common/simd/vec/operators.h:148:22: error: ‘__builtin_s390_vec_eqv’ requires z14 or higher
148 | #define npyv_xnor_b8 vec_eqv
| ^~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src:283:21: note: in expansion of macro ‘npyv_xnor_b8’
283 | npyv_b8 c = npyv_@VOP@_b8(a, b);
| ^~~~~~~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src: In function ‘simd_binary_less_equal_b8’:
numpy/core/src/common/simd/vec/operators.h:147:21: error: ‘__builtin_s390_vec_orc’ requires z14 or higher
147 | #define npyv_orc_b8 vec_orc
| ^~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src:236:21: note: in expansion of macro ‘npyv_orc_b8’
236 | npyv_b8 c = npyv_@VOP@_b8(a, b);
| ^~~~~~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src: In function ‘simd_binary_scalar1_less_equal_b8’:
numpy/core/src/common/simd/vec/operators.h:147:21: error: ‘__builtin_s390_vec_orc’ requires z14 or higher
147 | #define npyv_orc_b8 vec_orc
| ^~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src:260:21: note: in expansion of macro ‘npyv_orc_b8’
260 | npyv_b8 c = npyv_@VOP@_b8(a, b);
| ^~~~~~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src: In function ‘simd_binary_scalar2_less_equal_b8’:
numpy/core/src/common/simd/vec/operators.h:147:21: error: ‘__builtin_s390_vec_orc’ requires z14 or higher
147 | #define npyv_orc_b8 vec_orc
| ^~~~~~~
numpy/core/src/umath/loops_comparison.dispatch.c.src:283:21: note: in expansion of macro ‘npyv_orc_b8’
283 | npyv_b8 c = npyv_@VOP@_b8(a, b); |
seiko2plus
force-pushed
the
fix_ci_ibmz_assembler
branch
from
f800fca
to
d99a502
Compare
|
build passed after merges #21748 optimization log########### EXT COMPILER OPTIMIZATION ###########
INFO: Platform :
Architecture: s390x
Compiler : gcc
CPU baseline :
Requested : 'min'
Enabled : none
Flags : none
Extra checks: none
CPU dispatch :
Requested : 'max -xop -fma4'
Enabled : VX VXE VXE2
Generated :
:
VX :
Flags : -march=arch11 -mzvector
Extra checks: none
Detect : VX
: build/src.linux-s390x-3.8/numpy/core/src/multiarray/argfunc.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_unary_fp.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_arithm_fp.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_arithmetic.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_minmax.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_hyperbolic.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_comparison.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/_simd/_simd.dispatch.c
:
VXE : VX
Flags : -mzvector -march=arch12
Extra checks: none
Detect : VXE
: build/src.linux-s390x-3.8/numpy/core/src/multiarray/argfunc.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_unary_fp.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_arithm_fp.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_minmax.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_trigonometric.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_hyperbolic.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_comparison.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/_simd/_simd.dispatch.c
:
VXE2 : VX VXE
Flags : -mzvector -march=arch13
Extra checks: none
Detect : VXE2
: build/src.linux-s390x-3.8/numpy/core/src/umath/loops_trigonometric.dispatch.c
: build/src.linux-s390x-3.8/numpy/core/src/_simd/_simd.dispatch.c
|
|
@mattip, Any ideas to ask Travis to upgrade their images? the assembler needs to be upgraded to avoid incompatibility with gcc >= 9 |
|
reported to upstream https://travis-ci.community/t/s390x-upgrades-binutils-package/13083 |
|
Thanks @seiko2plus |
seiko2plus
added a commit
to seiko2plus/numpy
that referenced
this pull request
Jun 15, 2022
This guard protects against any sudden unexpected changes that may adversely
affect the compile-time SIMD features detection which could leave the SIMD code
inactivated. Until now we have faced two cases:
1. Hardening the compile-time test files of Neon/ASIMD features without checking
the sanity of the modification leading to disabling all optimizations on aarch64.
see numpygh-21747
2. A sudden compiler upgrades by CI side on s390x that causes conflicts with the
installed assembler leading to disabling the whole VX/E features, which made us
merge SIMD code without testing it. Later, it was discovered that this code
disrupted the NumPy build.
see numpygh-21750, numpygh-21748
seiko2plus
added a commit
to seiko2plus/numpy
that referenced
this pull request
Jun 15, 2022
This guard protects against any sudden unexpected changes that may adversely
affect the compile-time SIMD features detection which could leave the SIMD code
inactivated. Until now we have faced two cases:
1. Hardening the compile-time test files of Neon/ASIMD features without checking
the sanity of the modification leading to disabling all optimizations on aarch64.
see numpygh-21747
2. A sudden compiler upgrades by CI side on s390x that causes conflicts with the
installed assembler leading to disabling the whole VX/E features, which made us
merge SIMD code without testing it. Later, it was discovered that this code
disrupted the NumPy build.
see numpygh-21750, numpygh-21748
seiko2plus
added a commit
to seiko2plus/numpy
that referenced
this pull request
Jun 15, 2022
This guard protects against any sudden unexpected changes that may adversely
affect the compile-time SIMD features detection which could leave the SIMD code
inactivated. Until now we have faced two cases:
1. Hardening the compile-time test files of Neon/ASIMD features without checking
the sanity of the modification leading to disabling all optimizations on aarch64.
see numpygh-21747
2. A sudden compiler upgrades by CI side on s390x that causes conflicts with the
installed assembler leading to disabling the whole VX/E features, which made us
merge SIMD code without testing it. Later, it was discovered that this code
disrupted the NumPy build.
see numpygh-21750, numpygh-21748
rgommers
added
the
component: SIMD
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
On s390x, Travis CI runner fails to build SIMD VX features. The assembler errors seems to be caused due to invalid repo packages or incompatible assembler. the following error looks new, I didn't detect it before so maybe a new update has been made by Travis causes this issue.