This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH v2] Add math-inline benchmark
- From: OndÅej BÃlka <neleai at seznam dot cz>
- To: Carlos O'Donell <carlos at redhat dot com>
- Cc: Wilco Dijkstra <wdijkstr at arm dot com>, 'Siddhesh Poyarekar' <siddhesh at redhat dot com>, 'GNU C Library' <libc-alpha at sourceware dot org>
- Date: Thu, 23 Jul 2015 23:43:57 +0200
- Subject: Re: [PATCH v2] Add math-inline benchmark
- Authentication-results: sourceware.org; auth=none
- References: <55A907F6 dot 8000504 at redhat dot com> <20150717143406 dot GG19592 at spoyarek dot pnq dot redhat dot com> <002c01d0c30a$8f23c600$ad6b5200$ at com> <55AD4AC6 dot 5070401 at redhat dot com> <20150721113502 dot GA3415 at domone> <55AE7789 dot 2010704 at redhat dot com> <20150721173507 dot GA22893 at domone> <55AE9944 dot 3070406 at redhat dot com> <20150723151930 dot GA5009 at domone> <55B107B9 dot 5050209 at redhat dot com>
On Thu, Jul 23, 2015 at 11:26:49AM -0400, Carlos O'Donell wrote:
> On 07/23/2015 11:19 AM, OndÅej BÃlka wrote:
> >> If at all possible please provide actionable patches that show how you
> >> think the implementation should be.
> >>
> > I already send that previously in thread, read my benchmark.
>
> This is a long thread. Please help by providing a summary URL or reference
> directly to the full patch you show, that either applies on top of Wilco's
> work or replaces it as a complete math-inline benchmark.
>
There are two sets of benchmarks, for simplicity I first made this
standalone to show effects of inlining.
https://sourceware.org/ml/libc-alpha/2015-07/msg00343.html
Then I expanded my benchmark for following.
http://permalink.gmane.org/gmane.comp.lib.glibc.alpha/53314
These clearly show that there is no consensus about benchmarks yet as
these benchmarks give different results than Wilco and it isn't clear
what results are real and what are caused by bugs in gcc or selecting
specific workload.