Bizarre performance difference on G5x2.0, G5x2.5 and Intel

Apr 13, 2007 at 9:50pm

Bizarre performance difference on G5x2.0, G5x2.5 and Intel

This has me and the sysadmin scratching our head. A patch which renders some simple matrix transforms (reads from disk, does some simple pixel remapping and saves images) achieves about 19 fps on my personal dual 2.0 ghz G5, 1.5 gig ram, os 10.4.8.

The same patch gets around 12 fps at a lab on a dual 2.5 ghz G5 and 16 fps on a quad something Intel, both with at least 2 gig ram and macos 10.4.9. This makes no sense to me. Newer, higher spec machines would be faster, not slower.

I tried using different hard disks, different max/msp/jitter versions, running the pace fix for 10.4.9… But there is no significant difference. When using fw800 external raid0 disk I get 24 fps on my home machine and maybe 16 on the intel at the lab. Activity monitor shows that max uses between 12% and 18% cpu which is not much. No significant other processes. I’m stumped.

Anybody experienced this? Can this be a 10.4.9 issue, or is it something else?

All the best

Mateusz

#31368
Apr 13, 2007 at 10:35pm

L2 Cache differences?

Using openGL? Graphics card differences?

Lab machines with fragmented drives?

:)

On Apr 13, 2007, at 5:50 PM, Mateusz Herczka wrote:

>
> This has me and the sysadmin scratching our head. A patch which
> renders some simple matrix transforms (reads from disk, does some
> simple pixel remapping and saves images) achieves about 19 fps on
> my personal dual 2.0 ghz G5, 1.5 gig ram, os 10.4.8.
>
> The same patch gets around 12 fps at a lab on a dual 2.5 ghz G5 and
> 16 fps on a quad something Intel, both with at least 2 gig ram and
> macos 10.4.9. This makes no sense to me. Newer, higher spec
> machines would be faster, not slower.
>
> I tried using different hard disks, different max/msp/jitter
> versions, running the pace fix for 10.4.9… But there is no
> significant difference. When using fw800 external raid0 disk I get
> 24 fps on my home machine and maybe 16 on the intel at the lab.
> Activity monitor shows that max uses between 12% and 18% cpu which
> is not much. No significant other processes. I’m stumped.
>
> Anybody experienced this? Can this be a 10.4.9 issue, or is it
> something else?
>
> All the best
>
> Mateusz

v a d e //

http://www.vade.info
abstrakt.vade.info

#101814
Apr 14, 2007 at 8:39am

On 13 Apr 2007, at 23:35, vade wrote:

> Using openGL? Graphics card differences?

Doing straight Max patching with UI objects, my 500MHz G3 Pismo
PowerBook got a higher FPS figure than a 1.6GHz G5 tower.

– N.

nick rothwell — composition, systems, performance — http://
http://www.cassiel.com

#101815

You must be logged in to reply to this topic.