This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH v1.1 1/3] Copy over string performance tests - take 2
- From: OndÅej BÃlka <neleai at seznam dot cz>
- To: Siddhesh Poyarekar <siddhesh at redhat dot com>
- Cc: Andreas Jaeger <aj at suse dot com>, libc-alpha at sourceware dot org
- Date: Tue, 11 Jun 2013 08:24:18 +0200
- Subject: Re: [PATCH v1.1 1/3] Copy over string performance tests - take 2
- References: <20130604132207 dot GI13968 at spoyarek dot pnq dot redhat dot com> <51B4E0E5 dot 2060902 at suse dot com> <20130610053039 dot GA1570 at spoyarek dot pnq dot redhat dot com>
On Mon, Jun 10, 2013 at 11:00:39AM +0530, Siddhesh Poyarekar wrote:
> On Sun, Jun 09, 2013 at 10:09:09PM +0200, Andreas Jaeger wrote:
> > On 06/04/2013 03:22 PM, Siddhesh Poyarekar wrote:
> > >Hi,
> > >
> > >I had held back my earlier patch to start moving string performance
> > >tests from string/test-* to benchtests because Ondrej expresed
> > >interest in writing better tests. We're close to a freeze and I
> > >didn't see any submissions from Ondrej, so I've decided to revive this
> > >patch so that we at least have a starting point for string performance
> > >tests.
Are you aware that you are doing pointless labor here?
These benchmarks are about as far from real benchmark as possible(lack
of randomization, lack of testing with instruction cache testing over
nonrepresentative set of data, do not measuring relevant quantities,
not accounting for context switches etc.)
When somebody will try to use this benchmark he will write code that is
hideously wrong but maximizes number that only loosely resembles
performance.
Then we will need to fix again errors introduced by this and as it is
easy to avoid these pitfalls it is better not allow this possibility
in first place.