-
Notifications
You must be signed in to change notification settings - Fork 7.9k
Small change SIMD codes #18626
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.
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
base: master
Are you sure you want to change the base?
Small change SIMD codes #18626
Conversation
62e4508
to
253236d
Compare
@SakiTakamachi Thank you very much for update SIMD code. We need to confirm improve performance in aarch64 devices. I will try test on aarch64. Just a moment, please. cc @alexdowad |
Please check: #11076 that I was try on using NEON. Hmm... seems slow down in UTF-8 validation on Raspberry Pi 4B. configure
CPU
Using XSSE(This PR)
|
Maybe my environment is slow that added |
I used your benchmark code to run measurements on my M2 environment. It appears that the code utilizing NEON performs approximately 1.7 times better. Please note that the M2 has very strong SIMD capabilities, so the results might be better than what you would typically see on a standard ARM machine.
|
Thanks very much! |
I confirmed improve performance when release build. XSSE
non-SIMD(master dfff6ac)
Seems 2x faster than non-SIMD on Raspberry Pi 4B. |
That convert the SSE2 API to NEON using
zend_simd.h
, enabling the use of SIMD in a NEON environment.