Add non-temporal memcpy support for ARM #361
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
Update to use BOOST_TEST where possible instead of BOOST_CHECK_EQUAL or BOOST_CHECK. BOOST_TEST is more "modern" but doesn't cope with expressions that aren't be formatted with ostream, so it's not applied everywhere. The main motivation for this is that BOOST_CHECK_EQUAL leads to useless clang error messages when something goes wrong (it never reports the assertion that is the problem). This has been showing up with -Wsign-compare, and in some places the RHS of comparisons has been made explicitly unsigned to address these errors.
Previously -Dauto_features was used, but with the addition of SVE, no platform can have all features enabled. Instead, pass flags to force-enable features that we want to ensure are enabled on a per-platform basis. These are abstracted into a helper script to avoid repeating them for each usage. This script also takes care of -Dwerror and --native.
The implementation is very specific to x86, and naming it appropriately will make it easier to write an ARM version.
It's not actually used yet.
This is probably not optimal, but will provide something to test with.
It hurts performance and I'm confident it isn't going to be needed in any practical situation. Instead, update the documentation to indicate the corner case where the user might need to update code.
- Align the source address - Unroll the loop
This code path was only compiled if SVE is enabled and the ifunc attribute is not. Normally !ifunc is tested by the MacOS builds, but SVE is disabled in that case.
james-smith-za
approved these changes
Nov 7, 2024
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.
It uses SVE, so it'll run on Neoverse CPUs like Grace, but doesn't support any Apple silicon. We're not using Apple for performance anyway (just to enable Apple users to develop) so at this stage I'm not proposing to implement alternative paths for non-SVE capable machines.