This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH v2] Fix strrchr regression.
- From: OndÅej BÃlka <neleai at seznam dot cz>
- To: Liubov Dmitrieva <liubov dot dmitrieva at gmail dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Fri, 16 Aug 2013 13:06:27 +0200
- Subject: Re: [PATCH v2] Fix strrchr regression.
- References: <20130805173346 dot GA4978 at domone dot kolej dot mff dot cuni dot cz> <20130806083430 dot GA6122 at domone dot kolej dot mff dot cuni dot cz>
On Tue, Aug 06, 2013 at 10:34:30AM +0200, OndÅej BÃlka wrote:
Hi, according to profiling feedback I tuned strrchr implementation to
much faster one.
See results at:
http://kam.mff.cuni.cz/~ondra/benchmark_string/strrchr_profile.html
are updated and benchmark program is here:
http://kam.mff.cuni.cz/~ondra/benchmark_string/strrchr_profile160813.tar.bz2
I do not cover adding avx2 implementation which will be posted
separately. A considerable speedup is possible there.
http://kam.mff.cuni.cz/~ondra/benchmark_string/haswell/strrchr_profile/results_rand/result.html
I need to retest it on atom and silvermont to verify how much I
improved.
Passes tests, OK to commit?
Ondra
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Update.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c: Remove strrchr ifunc.
* sysdeps/x86_64/multiarch/strend-sse4.S Remove.
* sysdeps/x86_64/multiarch/strrchr-sse2-no-bsf.S Likewise.
* sysdeps/x86_64/multiarch/strrchr.S: Likewise.
* sysdeps/x86_64/strrchr.S (strrchr): Use optimized implementation.
---