P5W - FSB Wall testing and conclusion

[eva2000]

I'm on Intel D975XBX bad axe (no volt mods) max in bios is +50% on 1067fsb strap = boot at 9x400fsb and use clockgen to go higher heh

[mackanz]

Getting hard freeze at 420 with clockgen. Gotta try from 400 and see how high it will go.

[Counter CS]

I'm on Intel D975XBX bad axe (no volt mods) max in bios is +50% on 1067fsb strap = boot at 9x400fsb and use clockgen to go higher heh

Could it be BUS Speed limited to 1600Mhz ?
I see many people have a limit reached at 8x400 or 9x400Mhz BIOS limited, then to go higher you need some tweaks

Someone can confirm that ?

[eva2000]

nah it's all in the MCH it seems for 975x chipsets, 3 ways for higher fsb

1. more mch volts
2. loosening timings such as tRD and tRFC
3. better mch/nb cooling

MCH = memory controller hub

Explains why certain memory can take fsb higher on my bad axes i.e. 2x512MB Corsair PC5400ULv1.5 took me to max 421-423fsb on my 1st bad axe while 2x512MB OCZ PC5400 Plat XTC 3-3-3-8 only took me to max 378fsb. Different memory will handle tRD and tRFC values differently effecting max fsb ?

so makes sense really

[Counter CS]

Can someone talk to Asus guys to modify those BIOS to let us change those timings ?

[eva2000]

already forwarded the info on to my Asus contact.. my P5W DH arrived today as well

these memset options really do make a diff for dual stressprime testing and other stability tests!

[Wrangler88]

now when u change the settings in memset and reboot, are those settings saved or do u have to be able to get into windows, change in memset and then its stable?

[eva2000]

each windows session needs memset set again

[fhpchris]

do those two values effect real world memory performance ? or is it one of those settings that really only effects stability ?

hhmm odd i do not have tRD on my 965 chipset with the memset i use ? hhmm

Yes, those settings DO change performance, and tRFC of 40 is HIGH!

You should run subtimings closer to these, ignore speed as my T2600 walls here.

[Brahmzy]

Kind of a pain to have to reset memset every reboot on my daily box. Hmmmmmmmm.

[CrazyJohn]

i've done what's described in the 1st post and lost ~200MB/s in Goldmem...

but this suggests that bios timings are tight and if loosen better o/c can be done...

for Asus/Intel users:
can u manually set bootstrap? on DS3 there is no such option, nor to check it...

and about mem timings @ the same frequency, read this:
http://www.bleedinedge.com/forum/showthread.php?t=22297
basically - lower the fsb and higher the multiplier and you'll get better (tighter?) timings as the NB itself is working in a different mode than with high fsb

[Plaicd]

for Asus/Intel users:
can u manually set bootstrap? in DS3 there is no such option, nor to check it...

Not on Asus P5W-dh

[CrazyJohn]

so i have just tested and nothing...
rock stable @ 400Mhz... boots at 415 - and this is the last reasonable fsb to check, rest is not stable at all...

[Pjoeloe]

Kind of a pain to have to reset memset every reboot on my daily box. Hmmmmmmmm.

well , this prog isn`t created for 24/7 day use I gues. You can see what is limiting your mhz wall with this: the board ( bios ) , the memory or the cpu, so, pretty usefull I think.

[syne_24]

It didnt do nothing for me either. Still a wall at ~420

[Shark-357]

Its quite interesting i can do 7x460@1.75V VMCH but cant pass 9x423 even @1.85V VMCH.....

[Brahmzy]

Somethin' 'bout that 9x multi, eh?

[Pjoeloe]

Its quite interesting i can do 7x460@1.75V VMCH but cant pass 9x423 even @1.85V VMCH.....

Hmm, maybe the 9 x multiplyer is a little buggy in the C1 stepping of the p965 chipset for retail conroes "the C-2 Stepping is suppose to fix some FSB issues as well as some more optimizations on Intel's Fast Memory Architecture. Unfortunately C-2 Stepping P965 boards will not come till Sept. 9, 2006 "

this explains things in the Anand roundup, the 965s couldn't reach as high as the more mature 975 boards. Sounds logical because the ES samples were their, and with those chips the p965 C1 chipset is made.

Shark, did you do what`s on page one , first post ? did it help for your "9" multiplyer fsb ? 9 x 423 = 3807 mhz, maybe it`s the limit of your cpu

[Tony]

So now we have bios tweaking strap settings and also setting timings to tight for the ram to clock unless you have old fatty D9 which can stand it tight...ohh man there is no way we will win.

Im just anoyed i did not notice this sooner, so thanks for posting the info as this opens up the argument for us...we need fully opened up bios files to stand a chance of the boards clocking like they should.

[eva2000]

yup Asus already opened up P5B Deluxe bios AFAIK, waiting on P5W DH and D975XBX Badaxe bioses hehe

[Pjoeloe]

yup Asus already opened up P5B Deluxe bios AFAIK, waiting on P5W DH and D975XBX Badaxe bioses hehe

what`s your source besides that your board may be a nice overclocker ?

[eva2000]

what you do you mean i just forwarded the info from this thread to Asus and Intel contacts and now you see P5B Deluxe with new bios with all additional timings shown...

so logical assumption is sooner or later P5W DH and D975XBX bioses would be coming out following suit

[Danger30Q]

The fact that more BIOS options are available in the P5B Deluxe is promising for availability in a new P5W DH BIOS.

I love having the higher FSB on the P5B Deluxe but I absolutely need the Crossfire ability of the P5W DH for gaming. If only they could make a mix of the 2 boards....

All I hope for is new mem settings available in new P5W DH BIOS.

[aggybong]

I just tried this on my P5B-D and nothing changed I was so hopeful!

[Pjoeloe]

what you do you mean i just forwarded the info from this thread to Asus and Intel contacts and now you see P5B Deluxe with new bios with all additional timings shown...

Can you use the bios from the deluxe for the non deluxe? If not, can you mail asus and intel and ask if they want to do the same for the non deluxe ? I wanna try that board . Thanks