This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PING] Re: String Functions for x86-64
- From: "H. J. Lu" <hjl at lucon dot org>
- To: Rene Rebe <rene at exactcode dot de>
- Cc: libc-alpha at sourceware dot org, Ulrich Drepper <drepper at redhat dot com>, "Menezes, Evandro" <evandro dot menezes at amd dot com>, "Meissner, Michael" <michael dot meissner at amd dot com>
- Date: Fri, 28 Jul 2006 11:25:34 -0700
- Subject: Re: [PING] Re: String Functions for x86-64
- References: <84EA05E2CA77634C82730353CBE3A8430564AA8A@SAUSEXMB1.amd.com> <448A1DE8.7090106@redhat.com> <200607281310.31536.rene@exactcode.de>
On Fri, Jul 28, 2006 at 01:10:31PM +0200, Rene Rebe wrote:
> Hi,
>
> On Saturday 10 June 2006 03:18, Ulrich Drepper wrote:
> > Menezes, Evandro wrote:
> > > Ahem, I withdrew STRCMP, strncmp and strlen for now because their
> > > performance was similar to the current ones using GLIBC's tests.
> >
> > Because you apparently didn't managed to do so, I've contacted people at
> > Intel so that a blended implementation can finally be written. I expect
> > somebody will pop up sometime soon. Until this lead to some results
> > nothing is going to happen.
>
> Ping: Ullrich, any news about the mysterious "blended" codedrop?
>
> Maybe you can clarify what you do not like about the patch and what "blended"
> implementation should express?
>
We took a closer look at the new memcpy. We found that the new memcpy
was faster than the old one for sizes 1-1024 bytes, but much slower
than the old one for sizes > 1024 bytes. We are still investigating.
H.J.