On Jan 30, 5:46 pm, William Stein <wst...@gmail.com> wrote:
> On Fri, Jan 30, 2009 at 5:44 PM, mabshoff <mabsh...@googlemail.com> wrote:
>
> > On Jan 30, 2:18 pm, mabshoff <mabsh...@googlemail.com> wrote:
>
> > <SNIP>
>
> >> Yep, I have been playing around with this and I don't think it will be
> >> too far into the future.
>
> > I did play around with the script I wrote some more and the only code
> > using SSE3 or higher is
>
> > ./libatlas.so: at least 863330 sse instuctions used
> > found SSE3 haddpd: 1224
> > found SSE3 haddps: 530
> > found SSE3 movddup: 4
> > found SSE3 movshdup: 2
> > found SSE3 movsldup: 3
>
> > To make ATLAS use SSE2 only takes about 2 minutes in its spkg-
> > install.
>
> > Thoughts?
>
> SWEET!!!
I like to waste my days with pointless things like those :)
> This matches a lot with my experience, where one only ever notices the
> problems when moving binaries around when doing linear algebra.
>
> It would also be nice to have an easy option to build a binary that
> doesn't use sse2 either.
Hmm, do you mean SSE only, i.e. no SSE2 or higher or no SSE at all. No
SSE2 or high might be doable once I revamp the build system via #2999-
#3001 by tightly controlling CFLAGS and friends. But honestly while
doing #2999-#3001 is a medium priority at the moment doing a SSE only
build of Sage does not even show up on my radar yet since that implies
original Athlon A for example and I would think you need to look into
some hardware museum to find those in action :). To quote wikipedia:
[quote]
The following CPUs implemented IA-32 after SSE2 was developed, but did
not implement SSE2:
* AMD CPUs prior to Athlon 64, including all Socket A-based CPUs
* Intel CPUs prior to Pentium 4
* Via C3
* Transmeta Crusoe
[end quote]
Seriously: The energy spend on an SSE only build is plainly not worth
it considering the number of bugs I can fix in the same time where the
vast majority of Sage users benefit. Once #2999-#3001 is done we can
attempt to do it.
> -- William
Cheers,
Michael
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
