B4 that i was using windsor 3800+ ...same problem also , but that 3800+ cannot push further anymore so i thought it is bottlenecked by the proc . After swap to brisbane, I've realized that actually it is chipset problem.

But look at this, Pi result of pcmoddingmy. He is using DFI Infinity NF UltraII-M2 (BIOS D711) , same as yours i think. Proc 3.1GHz and ram 1042 , he is definitely using 667 divider during that time.



Click here for more detail : http://forum.lowyat.net/topic/321443/+20

oic ....i do not know that man. You got the point, maybe it is bcoz of brisbane's crappy divider also. PCmoddingmy was using an athlon64 3000+ during that time, a 90nm proc.

and 1 more , Brisbane has higher L2 cache latency, that's why it runs slower than windsor in SPi.

But got pros also, can OC high. I've just realized that 3.25GHz is not the limit of my proc yet... can push a little bit further. Hope can enter Pi 1M 26sec during next try

This post has been edited by dblooi: Jul 23 2007, 11:14 AM

1.6vcore

someone has told me that he've got a nice stepping brisbane , if cannot get CZ maybe i will ask him whether can borrow me his brisbane for one night or not

haha, yeah just for PI running. The rest of the time i'm using it at stock speed and stock voltage.

I dunno yet... It's not good to give a wild prediction here . I will try it in this weekend

stock voltage can hit 300X9.5 , not an insane stepping la ....

Prime? I haven't try. Will try it together in this weekend as well...

This is the PI screenshot , taken when i was using TF6100AM2.

300X9.5 stock vcore




Attached thumbnail(s)

wow you serious?

i try to push my brisbane during this weekend 1st, if cannot then i'll find you ok?

I've read your signature. What's the vcore that u pumped in order to get 2.8GHz? Or issit stock vcore?

This post has been edited by dblooi: Jul 23 2007, 10:51 PM

ooooo..... by looking at it, it seems has very big potential to do a big job....

This post has been edited by dblooi: Jul 23 2007, 11:02 PM

his one is CZ , hitting 3GHz is not a big deal.....

if 5200 above nvm liao , don't feel like wanna spend so much on the cpu. Let me think 1st ok. Btw, i think 5200 above confirm all CZ 1 .

Anyone here know about the Super PI OS tweak -- copywaza?

I've tried to perform it but i failed due to unknown reason. However , the PI result already got big improvement.

b4 the tweak , 29.984s.



after the failed tweak , 29.5s



anyone know the correct way to perform copywaza?

btw, the result has improvement maybe bcoz I'm using slightly different timing setting for each test.. but the possibility is not high as i've tried many times b4, not much different between these 2 set of ram timing

This post has been edited by dblooi: Jul 24 2007, 10:14 AM

haha issit? I don't think so. Anyway, i will try and see. Thanks for your comment

as you wish

take a look at the memset sub-timing. I even adjust them for the best performance. Useless, still slower than my previous result

edited: i din tune the subtiming MAL as my pc will hang sometimes once i adjust it. Suppose it should be adjusted .

This post has been edited by dblooi: Jul 24 2007, 04:29 PM


Attached thumbnail(s)

I'm using an integrated graphic mobo (no money to buy gc ). The Pi result might be slightly slower than others

haha. Thanks alot. Btw, you can take a look at this article ( if you can read chinese).

http://bbs.ocer.net/viewthread.php?tid=138056&highlight=

It shows the relationship between those ram timing and usage of them. The author is OPB , an OC master from taiwan. He still got a lot of articles that teach people how to set ram timing and some OS tweak in order to run PI faster. Can try to google it

TF560 AM2+ is just another am2 mobo, max 1000 HT link as well.

HT link is OCable , i saw people push it to 1500 b4

well ...i have no idea. As i know, OC HT link will improve the performance of integrated gc , that's why people did that. But the case that u mentioned is more like oc the PCI-E and PCI clock.

that is just an advertisement from Biostar. Real mobo only got one 4pin and the mosfet of each phase has been reduced to 3



This post has been edited by dblooi: Jul 24 2007, 11:42 PM


Attached thumbnail(s)