hyperscan: support aarch64 platform #212
Open
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
tqltech
force-pushed
the
aarch64
branch
2 times, most recently
from
c039c90 to
dfd4b80
Compare
|
Arm has invested in adding aarch64 support with x86-specific code abstracted away and arch specific code put into separate headers in src/util/arch/. This way, maintaining arch-specific code of hyperscan and/or porting hyperscan to a new architecture (eg Power/RISC-V/etc) should now be much easier. We have all tests passing for both architectures, and with our next step being to add NEON SIMD optimizations for aarch64. We would appreciate your review/comments of our approach here: #286 |
Closed
|
aarch64 support exists here: https://github.com/VectorCamp/vectorscan |
Closed
1 task
cosmicexplorer
pushed a commit
to cosmicexplorer/hyperscan
that referenced
this pull request
Dec 30, 2023
SIMDe on Clang needs SIMDE_NO_CHECK_IMMEDIATE_CONSTANT defined and other SIMDe related fixes now that SIMDe is part of the CI pipeline. Some issue with SIMDe on x86 still remains because of an upstream bug: simd-everywhere/simde#1119 Similarly SIMDe native with clang on Arm also poses a non-high priority build failure: https://buildbot-ci.vectorcamp.gr/#/builders/129/builds/11 Possibly a SIMDe issue as well, need to investigate but will merge this PR as these are non-blockers.
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Outdated suggestions cannot be applied.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.
Add aarch64 platform branch, which is fully compatible with armv8-a, and at the same time ensure that the use of x86 platform is not affected.
The modification can be divided into the following aspects :
(1) CMakeLists.txt: add branches to judge x86 and aaarch64 platforms.
(2) simd_arm.h: Use the neon instruction to implement 128 operation functions.
(3) simd_x86.h: Save the original simd_utils.h file contents.
(4) simd_utils.h: Select different header files (simd_x86.h or simd_arm.h) according to the platform judgment results.
(5) Intrinsic function: There are multiple calls to the intrinsic function in the code, and the intrinsic function is encapsulated in the modification.
(6) control_verbs.cpp / ExpressionParser.cpp / control_verbs.cpp / Parser.cpp: There are errors in generating these four files under the aarch64 platform. So the files are generated on the x86 platform and copied to the corresponding directory for use. Does not affect the functionality of the x86 platform.
On the basis of support the aarch64 platform, we also carried out a large performance optimization. The optimization effect also improved the performance of hyperscan on the x86 platform. The currently submitted PR is a successful migration version, and we expect to submit a performance version later.