PII for team 32

[jonspd]

Just wondering am I the 1st one to be folding on a PII for team 32? :D


some improvement over my 9850be thus far hasn't been maxed out or fully tested at all.


jonspd

[KonaKona]

No I think I had a 500mhz stick folding for a while, but then I decided it was better for throwing off high places than folding. :)

Where are everyone getting these? Though they haven't been released yet?

[sno.lcn]

Hehe he means Phenom II :beer:




*edit: I'm an idiot, didn't finish reading the second half of your post :p

[Cluster]

What kind of PPD are you seeing with it? Would be sweet to see these things breaking 10m frame times.

[KonaKona]

Hehe he means Phenom II :beer:




*edit: I'm an idiot, didn't finish reading the second half of your post :p

No no, I thought intel was still sitting on those kalamaths!

I'd be nice to see that AMD found a way to get 20k PPD out of a processor, or something outrageous like that. But it would also be nice to see newegg selling PIIs. Last I checked they were supposed to be on sale 12/22 but apparently that didn't happen.

[jonspd]

What kind of PPD are you seeing with it? Would be sweet to see these things breaking 10m frame times.

not gonna be less the 10 mins IMHO from the looks of it I get about 1700ppd on a 2665 at 3.6ish

[jonspd]

cluster just for you

q9450 @ 3.4 = 2890 ppd on 2653
PII @ 3.5 = 2444 ppd on 2653

numbers from F@Hmon

[dfonda]

cluster just for you

Better hurry before Cluster buys them all!:p:D

[MRD]

Hmm, I'm at 8-9 mins/percent on a Phenom 9950 @ 2.6 running the SMP Linux client. It's doing about 3k ppd. Is that unusual? I just started it this week... went from blue to dark red.

[Cluster]

cluster just for you

q9450 @ 3.4 = 2890 ppd on 2653
PII @ 3.5 = 2444 ppd on 2653

numbers from F@Hmon

Im guessing thats windows smp. I get ~4000ppd on my Agena @ 2.5ghz :p

I guess i should have been more specific. I meant 10 minutes on a dual-core VM. Would push the Deneb closer to the 6k PPD mark. Still not holding a candle to the 9kppd i7s... friggin cheating HT :p

Hmm, I'm at 8-9 mins/percent on a Phenom 9950 @ 2.6 running the SMP Linux client. It's doing about 3k ppd. Is that unusual? I just started it this week... went from blue to dark red.

If thats native in linux, its not bad. Are you getting A1 or A2 cores? Collectively with 2 VMs through windows, each VM has 2 cores, they do 12m/frame each, effectively 6m/frame for the quad @ 3ghz. The VM in generally is pretty good at limiting overhead, so i dont think there's a big difference, although native linux shouldnt be worse.

What kernel version? I know there was a regression in one of the new kernel releases, i dont know the version off hand. Whichever comes with Ubuntu 8.10, there was something that made folding take a hit, a good 30% in some cases.

Oh, and jon, can you get me one :D

[jonspd]

yep window smp on XP, no vm's on my main rig anymore :D


there is only one left where I got it from and I think it is already taken.... just ask the #3 poster lol

[MRD]

I'm running Gentoo Linux 64 bit (multilib) kernel 2.6.25 (I think), gcc 4.3.2 with -march=amdfam10 set in cflags.

It's also my main rig so it spends a fair amount of cpu time doing other things too. Nothing virtual there though, it's just running Linux and folding in Linux. I could probably up the points if I didn't use it for everything else though.

Getting a2 cores... is that good or bad? What is the difference?

[Cluster]

If your using it lots than ya, thats pretty good.

A2 cores are the ones you want, they use a newer version of gromacs, 2.01, which is much more advanced in terms of SMP. It will utilize your CPU better. A1 core uses an older 1.9x gromacs and it really doesnt scale well. It can be more than twice as bad on a quad core as A2. The windows SMP only gets A1 cores. So you can see the difference, with jon getting ~2500ppd on his deneb at 3.5ghz, and me getting ~4800ppd on my agena at 3ghz,

[MRD]

What kernel version? I know there was a regression in one of the new kernel releases, i dont know the version off hand. Whichever comes with Ubuntu 8.10, there was something that made folding take a hit, a good 30% in some cases.

Is it the kernel itself or something else related to ubuntu 8.10? Has the slowdown been noted in other distros running the 2.6.27 kernel? Has the reason been determined? You could probably get around it with some kernel configuration.