Asus A8V

[Sierra_abx]

This post on another forum explains the difference between FSB and HTT. Technically, there is no FSB on A64 boards which use hypertransport instead. But in a loose sense, FSB and HTT are roughly equal.

[Riptide]

Thanks for the explanations Sierra Bound. I am going to try and tip my hat @ overclocking again but am waiting until I can get a different motherboard then this AV8. I'm not sure it's been the source of my problems yet or not but I suspect as much. When I can finally get my nforce3 I'll give it another try.

[Blue078]

SB on 939 the memory speed is directly tied to the multiplier ? Yes

ie" @240htt x9 would result in less membandwith than 240x10 or 240x11
am I some were close in my understanding ?

And do you think the BIOS would have any influence over this?

Thanks

[Sierra_abx]

Hi Blue. I'm not certain about the relationship of the multi's to bandwidth. I do know that the .5 multi's use a RAM divider that affects memory speed. So if, for instance, you use the 9.5 multi, you calculate the actual RAM speed using the next highest whole multi which in this case would be 10.

200 (RAM speed) X 9.5 (multi) = 1900
1900/10 = 190 actual speed

[maxxshwagg]

don't think i am going to replace my asus with a rev 2. i called them up and they said that i would have to ship them the board and they would test it before getting a replacement. since lock weren't promised then it doesn't qualify as a rma....anyways who wants to give up their board for 3-4 weeks. at least abit has a cross ship policy...

anyways

just ordered a msi neo2 nforce3 from new egg $149 (along with a 74 gig raptor). lets see how it compares to the asus.....

max

damn, now i have a A8V, IC7Max2, and a p4 3.0c (that hits 3.87) sitting on my shelf. guess i'll be listing some stuff on ebay...

[Blue078]

Originally Posted by sierra_bound

Hi Blue. I'm not certain about the relationship of the multi's to bandwidth. I do know that the .5 multi's use a RAM divider that affects memory speed. So if, for instance, you use the 9.5 multi, you calculate the actual RAM speed using the next highest whole multi which in this case would be 10.

200 (RAM speed) X 9.5 (multi) = 1900
1900/10 = 190 actual speed

I kinda noticed something on the SK8V, but it's real clear on the this board.

mem set at cas2.5-3-3-8-1t 1:1 {sync}, sandra 2004 SP2 memory bandwidth test.
250x8=... 6278/6126
250x9=... 6426/6348
250x10=. 6602/6534

250x11 isnt possable untill I this thing is frozen, but x11 is the 3500+'s default multi and this is were the chip produces it's best bandwidth.

What I'am getting at with all this is, I think this is were some people new to A64 are haveing some OC problems.

The Ballistix memory I have is perfectly stable @250x8 or 250x9 c2.5-3-3-8-1t
And is benchable @250x10 cas2.5-3-3-8-1t....
but during the course of 30 min's of 3D game play, crashes occur {250x10}.

Hold on before you say Blue it's your CPU 3D games are stable at 250x10, but only after raiseing the mem timings to 3-4-3-8-1t

?????????????????????

[xgman]

Originally Posted by sierra_bound

Hi Blue. I'm not certain about the relationship of the multi's to bandwidth. I do know that the .5 multi's use a RAM divider that affects memory speed. So if, for instance, you use the 9.5 multi, you calculate the actual RAM speed using the next highest whole multi which in this case would be 10.

200 (RAM speed) X 9.5 (multi) = 1900
1900/10 = 190 actual speed

huh? that doesn't make any sense. If I run at 200 x9.5 sync, my ram speed is still 200. my 1/2 multis work as they should on the FX53. maybe this applies to non-FX cpu's. If I test my ram in a dozen diffeent ways it is still an htt of 200 at the above settings.

[xgman]

My 2.0 is being delivered today.

[hackysak]

Originally Posted by xgman

My 2.0 is being delivered today.

xg,

Where'd you purchase your 2.0 from?

[Purple Lizzard]

Originally Posted by xgman

My 2.0 is being delivered today.

Good luck xgman! We can't wait to hear your results.

[:PopCorn]

xgman,
What Bios is included with R2.0? Are Locks available & working?

[Sierra_abx]

Originally Posted by xgman

huh? that doesn't make any sense. If I run at 200 x9.5 sync, my ram speed is still 200. my 1/2 multis work as they should on the FX53. maybe this applies to non-FX cpu's. If I test my ram in a dozen diffeent ways it is still an htt of 200 at the above settings.

Read this thread.

[Sierra_abx]

Originally Posted by Blue078

I kinda noticed something on the SK8V, but it's real clear on the this board.

mem set at cas2.5-3-3-8-1t 1:1 {sync}, sandra 2004 SP2 memory bandwidth test.
250x8=... 6278/6126
250x9=... 6426/6348
250x10=. 6602/6534

250x11 isnt possable untill I this thing is frozen, but x11 is the 3500+'s default multi and this is were the chip produces it's best bandwidth.

What I'am getting at with all this is, I think this is were some people new to A64 are haveing some OC problems.

The Ballistix memory I have is perfectly stable @250x8 or 250x9 c2.5-3-3-8-1t
And is benchable @250x10 cas2.5-3-3-8-1t....
but during the course of 30 min's of 3D game play, crashes occur {250x10}.

Hold on before you say Blue it's your CPU 3D games are stable at 250x10, but only after raiseing the mem timings to 3-4-3-8-1t

?????????????????????

Interesting tests you did there. I didn't realize the multi makes that much of a difference. When you say it's benchable at 250X10, are you taking about 3DMark or SuperPi? Generally, you can bench SuperPi at a higher speed than 3DMark.

[Blue078]

Originally Posted by sierra_bound

Interesting tests you did there. I didn't realize the multi makes that much of a difference. When you say it's benchable at 250X10, are you taking about 3DMark or SuperPi? Generally, you can bench SuperPi at a higher speed than 3DMark.

3D Mark 2001 will complete 5 loops with out fail, ans 3dM2003 runs with out issue. And TBH I didnt even think about trying super Pi. I found along time ago Bench stable and game stable are not allways one in the same

I'm trying to figure out what is a fault here, it's just not super clear to me whats at fault the mem, or the on die mem controller is faileing at higher htt's, and thats why raiseing the mem timings gain stability.

I guess the only way to sort this out is spend the time and effort, and freeze the chip..then and see if I can keep the tighter timings.

[Sierra_abx]

My board does not like dividers. I tried running 290X10 with a divider and almost hosed my OS. The board doesn't like HTT over 280.