Q9*40 and X33*0 C1 FSB Problems.

[RunawayPrisoner]

I thought as much. Usually, an overclock should be possible on all BIOSes, and it would stay stable on some specific BIOSes. It doesn't necessarily mean that a BIOS would take away an FSB limit. How many times has that (BIOS taking away FSB wall) happened, actually?

[jas420221]

So what is the max GTLvref volts then? I started my own thread a week ago asking this but only got 1 person to reply (who was helpful). For some reason, my BIOS sets GTLvref 1 @ 45mv while on auto (stock clocks, and overclocked it didnt change, but I knocked it back to 00mv manual for all). Im nervous to pump a lot of volts through that thing. I dont want to buy another $350 processor.

Here is my thread so as not to jack this one (just in case): http://www.xtremesystems.org/forums/...d.php?t=183841

[FiXeR]

So tweaking the GTL wil solve a little bit the FSB problem? Gonna try that when I'm home from school.

[RunawayPrisoner]

So tweaking the GTL wil solve a little bit the FSB problem? Gonna try that when I'm home from school.

Um... sorry to bust your bubbles... but I think it might not help at all. From the article I linked earlier, it seems GTL will only make an impact on stability, and not precisely on overclockability.

P.S.: But someone reported an FSB strap at 450 (which means 451 would work but 450 might not) so... maybe that's a clue.

[ChaosMinionX]

Yep, I was playing with board all day today..... the board booted better on Rampage Formula with 266mhz / 451mhz FSB... 450mhz/451mhz on 333/400mhz strap was a no go. I really think this is all a BIOS issue.

[FiXeR]

Could it be a Strap problem: http://www.xtremesystems.org/forums/...&postcount=281

[rx7speed]

goin to read this when I get home.

[RunawayPrisoner]

Seems like it IS a strap issue, guys. We have had people reporting 475 stable now thanks to this finding. Great job, ChaosMinionX!

[003]

I hope it IS a strap problem. Because let me be the first one to report, running on a DFI P35, my Q9450 has been absolutely awful at overclocking, cant get over 350 FSB. 400, 450, 475 all fail prime95 on two cores almost instantly at these speeds... and cranking voltages to the limit (1.45 vcore, 1.4vtt, 1.6NB, 2.4 dram) does not make a lick of difference. See my thread here:
http://csd.dficlub.org/forum/showthread.php?t=6336

[RunawayPrisoner]

It seems now that it's half a strap problem... and half not? And anyway, another member got the same board: ASUS P5E-VM HDMI, and immediately, he could POST at 475.
Myself, I've been able to POST at up to 524 now... and I think that's a bit... high?

[003]

I can POST at high FSB (475+). But, anything over 350 causes two cores to fail prime95 the second it is started, regardless of how much voltage I am feeding everything. This is pissing me off to no end because I am on extremley good water, my BIOS reports the CPU temp at 8-11C, but I can hardly get a 200MHz overclock!!!

There has to be something else at work, because the amount of time it takes to fail prime is always the same, regardless of speed or voltage.

[RunawayPrisoner]

Well, POSTing can be quite... fun.

Gee... but only 1MHz to 4200MHz and it wouldn't let me... so it ends up being 4192 instead. Oh well.

[FiXeR]

A guy from Gathering of Tweakers (A dutch forum) called to Asus and he talked to the BIOS Department fron Asus. He explained the problem to Asus and the will look what the problem is and he'll fix it! Next week we hear more from Asus and what we can do to fix the FSB limit on the X48/X38/P35 motherboards. Maybe they will come with a new bios with the fix in it.

Great support from Asus!

[003]

Well, POSTing can be quite... fun.

Gee... but only 1MHz to 4200MHz and it wouldn't let me... so it ends up being 4192 instead. Oh well.

Dude... thats really not helping the situation, bragging...

Fixer:
I am seriously considering getting the x48 asus board (ddr2 version), as it just seems like a higher quality board all around (better heatsinks, 8 layer pcb, etc...) and does not have retarded ram compatibility problems.

How is vdroop on it?

[FiXeR]

Dude... thats really not helping the situation, bragging...

Fixer:
I am seriously considering getting the x48 asus board (ddr2 version), as it just seems like a higher quality board all around (better heatsinks, 8 layer pcb, etc...) and does not have retarded ram compatibility problems.

How is vdroop on it?

I've a P5E flashed to a Rampage Formula. (Still a X38) But with the rampage bios do you have minumum vdroop. Dont know how much, but look around in the X48 topics

[RunawayPrisoner]

Dude... thats really not helping the situation, bragging...

Yeah, but what I'm trying to say is that this is not specific to ASUS boards. It also happens with Abit and DFI boards as well, as reported by a lot of others. Maybe it is a BIOS issue and a BIOS update for all X38/X48 boards is in order, but my suspicion up until now falls on the design of the chipset rather than the BIOSes. If that's true, then all X38/X48 boards might not help at all in this situation. But ASUS P5E-VM HDMI is not the only way to go. You can still choose the Striker II Extreme, and from what I've heard, that board could also hit 4GHz easily.

[ChaosMinionX]

Yeah, but what I'm trying to say is that this is not specific to ASUS boards. It also happens with Abit and DFI boards as well, as reported by a lot of others. Maybe it is a BIOS issue and a BIOS update for all X38/X48 boards is in order, but my suspicion up until now falls on the design of the chipset rather than the BIOSes. If that's true, then all X38/X48 boards might not help at all in this situation. But ASUS P5E-VM HDMI is not the only way to go. You can still choose the Striker II Extreme, and from what I've heard, that board could also hit 4GHz easily.

Its turning out to be a X38/X48/P35 issue with clocking these, most likely they didnt tune them properly for 45nm and focused more on 60nm which was the current market when all these boards were released...so if people could not clock 60nm, then nobody would buy them.

Now that 45nm is hitting the streets (slowly) but the issues are starting to come out, my guess is they never pushed these boards with 9450/3350 to 450-475mhz FSB. Part of that would explain why DFI's boards work with higher FSB, because they do extreme testing like that from what I have seen.

[RunawayPrisoner]

But I'm also getting that DFI boards are not clocking as well either despite the user having total control over almost anything adjustable. Currently, world record is held by a P5K modded to death (with very insane voltages) so I believe that with even higher voltages on these boards, it's possible for the chip to do 450 - 475, but if that's so then it's a board design issue and not a BIOS issue. Most likely these boards have been designed to clock 65nm processors in mind, and not to clock 45nm processors.

But then I can't explain why the P5E-VM HDMI requires such low voltages to do extreme clocks. Here's what I need for 500MHz:

vCore: 1.42v real (1.48v under load if the system is to stay stable)
vPLL: 1.58v - 1.60v
vFSB: 1.36v - 1.40v
vNB: 1.65v real (bare minimum)

So 500MHz doesn't require much effort from the board... if it required any at all.

Edit: the said P5K results:

http://www.xtremesystems.org/forums/...d.php?t=182791

Compare that to the kind of voltages I need for 500MHz (stable) and you'll see how much of a difference it makes. Maybe vCore stays the same, and I'll need more than 1.6v to validate that 524MHz run (or to POST further than 524MHz) but other voltages for the P5K board were really high, almost impractical for assuring longevity.

[003]

Part of that would explain why DFI's boards work with higher FSB, because they do extreme testing like that from what I have seen.

Not faster with my experience. See my previous posts in this thread.

Hmm that P5k seems to do really well, I need to look into that.