P35 Double Boot

[toadeater]

My EP35-DS3R is suffering from the infamous double-boot problem. It reverts to stock settings after every cold boot if I OC higher than ~3.4GHz. It's 100% stable in Windows at 3.8GHz, tested with games, prime95, video, multitasking, etc. If I reboot it maintains the OC, but if I shut the power down it will double-boot.

What I tried so far:

Lots of different voltage settings. It's stable in Windows, so I don't think that's the problem.

Auto and manual RAM timing. Apparently didn't make any difference. Even with the ram underclocked (375x9.5 instead of 400x9) it still double-boots.

Flashed to BIOS F3e. That locked my CPU voltage to stock and I couldn't change it at all.

Flashed back to F2, could change voltages again, but still get double boots.

Cleared CMOS (removed battery, power, etc) numerous times already.

Disabled legacy USB support and halt on errors. Made no difference.

Disabled C1E, EIST, etc.-- again it made no difference. They actually work fine in Windows at 3.8GHz so I turned them back on, no BSODs.


So has anyone ever solved this kind of thing on their board or have any ideas what to try next?

 

[VirtualLarry]

Do you have any external USB devices? Are they attached when you attempt to boot?

 

[Heidfirst]

Originally posted by: toadeater
but if I shut the power down it will double-boot.

by shut the power down do you mean just let Windows shut it down to a "soft off" status or do you physically switch off power either at the PSU or wall socket?

 

[thegisguy]

My Abit IP35-E only double posts when I unplug it from the wall, or switch off the power supply. Otherwise under normal use it's fine. So you unplugging it like Heidfirst asked?

 

[Budarow]

Originally posted by: toadeater
My EP35-DS3R is suffering from the infamous double-boot problem. It reverts to stock settings after every cold boot if I OC higher than ~3.4GHz. It's 100% stable in Windows at 3.8GHz, tested with games, prime95, video, multitasking, etc. If I reboot it maintains the OC, but if I shut the power down it will double-boot.

What I tried so far:

Lots of different voltage settings. It's stable in Windows, so I don't think that's the problem.

Auto and manual RAM timing. Apparently didn't make any difference. Even with the ram underclocked (375x9.5 instead of 400x9) it still double-boots.

Flashed to BIOS F3e. That locked my CPU voltage to stock and I couldn't change it at all.

Flashed back to F2, could change voltages again, but still get double boots.

Cleared CMOS (removed battery, power, etc) numerous times already.

Disabled legacy USB support and halt on errors. Made no difference.

Disabled C1E, EIST, etc.-- again it made no difference. They actually work fine in Windows at 3.8GHz so I turned them back on, no BSODs.


So has anyone ever solved this kind of thing on their board or have any ideas what to try next?

I thought this was a common problem for ABIT's IP35 mobos which was solved in a BIOS update. Might be the same for your mobo maker? Maybe a slightly earlier BIOS version from your mobo maker will help?

 

[toadeater]

Ok, I think I found the problem. I've gone through two boots now, 400x9.5 has stuck even though 365x9.5 wasn't even booting before. What I did this time was raise the MCH voltage to +2v and that seems to have done the trick. I hadn't done this earlier because when I could boot, it was stable in Windows. I thought it might have been the CPU voltage, but I could be 100% stable in Windows with a CPU voltage of ~1.3v @ 400x9.5 so it didn't make sense why it couldn't even boot at 365x9.5.

http://img504.imageshack.us/im.../6735/overclockay1.jpg

This has to be a BIOS problem of some kind, because there's no need for such high voltages except to get this board to boot. This is on an EP35-DS3R, btw. It seems to be a frequent problem on Gigabyte boards lately. I wonder if DES has anything to do with it?

 

[toadeater]

Nope, back to double booting again after I left the PC turned off for a few hours. Several failed boot attempts later it finally boots into Windows, 100% stable again. Doesn't make any sense.

 

[toadeater]

Update:

I've continued to try various other things to get this board to cold boot and retain OC settings. Usually, I have been able to get back up to 3.6 or 3.8GHz after 3-4 reboots (reboot, load BIOS settings, reboot... ), and increasing the voltages to +2 for MCH and FSB worked a couple of times. I also tried manually setting all the RAM sub-timings, which made no difference. I didn't think it would because I'm not OCing the RAM at all, something else is causing this. Voltages didn't help, RAM timings didn't help, WTF is it?

After more research I noticed that Gigabyte boards were having similar problems waking from S3 sleep, In those threads they suggested to enable support for USB keyboards and mice, and set the first graphics adapter to PEG (VGA) instead of PCI. I tried it and it surprisingly worked today; booted to 3.6GHz @ 1.25v in one try. Probably just a fluke, but who knows?

Too early to tell yet if it's a permanent fix, I'll update this thread again.

 

[Budarow]

Originally posted by: toadeater
Update:

I've continued to try various other things to get this board to cold boot and retain OC settings. Usually, I have been able to get back up to 3.6 or 3.8GHz after 3-4 reboots (reboot, load BIOS settings, reboot... ), and increasing the voltages to +2 for MCH and FSB worked a couple of times. I also tried manually setting all the RAM sub-timings, which made no difference. I didn't think it would because I'm not OCing the RAM at all, something else is causing this. Voltages didn't help, RAM timings didn't help, WTF is it?

After more research I noticed that Gigabyte boards were having similar problems waking from S3 sleep, In those threads they suggested to enable support for USB keyboards and mice, and set the first graphics adapter to PEG (VGA) instead of PCI. I tried it and it surprisingly worked today; booted to 3.6GHz @ 1.25v in one try. Probably just a fluke, but who knows?

Too early to tell yet if it's a permanent fix, I'll update this thread again.

Please do update this thread

 

[toadeater]

Ok, the latest attempt didn't work either. It worked once, then it went back to the usual double-boots. The cold boot problem starts somewhere after ~367 FSB, because I tried 375 and 369 and they both failed on some cold boots. It seems to be FSB related, since if I try 400x8.5, it will fail to boot, but 367x9.5 will boot. I tried as much as +.2v on the MCH and FSB, along with very loose RAM timings, and even that didn't make any difference. It doesn't seem to be related to actual stability, but something else. Maybe it's my PSU after all, but why would it fail only on cold boots?

I'm most annoyed that my E7200 can hit at least 3.8GHz stable on 1.296v on this board, but the BIOS error, or whatever it is, is limiting me to 3.5GHz unless I feel like rebooting enough times until the OC finally sticks.

 

[Bigs11]

Originally posted by: toadeater
Ok, the latest attempt didn't work either. It worked once, then it went back to the usual double-boots. The cold boot problem starts somewhere after ~367 FSB, because I tried 375 and 369 and they both failed on some cold boots. It seems to be FSB related, since if I try 400x8.5, it will fail to boot, but 367x9.5 will boot. I tried as much as +.2v on the MCH and FSB, along with very loose RAM timings, and even that didn't make any difference. It doesn't seem to be related to actual stability, but something else. Maybe it's my PSU after all, but why would it fail only on cold boots?

I'm most annoyed that my E7200 can hit at least 3.8GHz stable on 1.296v on this board, but the BIOS error, or whatever it is, is limiting me to 3.5GHz unless I feel like rebooting enough times until the OC finally sticks.

I'm having exactly the same problem as you with my E6750 @ 3.6Ghz (450x8) running on GA-P31_S3L board.
I'm trying extra FSB (+.2v) and MCH (+.1v) at the moment, as it cold booted today with those settings.
Sometimes I think I'm onto it, but then other times I need to increase the Vcore to 1.45v to make it hold.
Other times, it'll be completely stable with a 1.36v Vcore.
Which also leads me think it might be PSU. Mine is a 430W PSU, which I thought would be plenty, as I'm only running a stock Albatron 7300 Vid Card.
Will follow your thread with interest in case you crack the problem.

 

[toadeater]

Originally posted by: Bigs11
I'm having exactly the same problem as you with my E6750 @ 3.6Ghz (450x8) running on GA-P31_S3L board.
I'm trying extra FSB (+.2v) and MCH (+.1v) at the moment, as it cold booted today with those settings.
Sometimes I think I'm onto it, but then other times I need to increase the Vcore to 1.45v to make it hold.
Other times, it'll be completely stable with a 1.36v Vcore.
Which also leads me think it might be PSU. Mine is a 430W PSU, which I thought would be plenty, as I'm only running a stock Albatron 7300 Vid Card.
Will follow your thread with interest in case you crack the problem.

Yes, that's it. I can cold boot sometimes, but the settings that work vary from day today. Yesterday 365 FSB worked, today I had to drop it to 364 FSB. If I reboot a few times, I can get 400 FSB to work, until the next time. I have a 550W PSU, voltages are steady.

The only thing left I can think of is that I have fans attached to all 5 motherboard fan headers, which might be causing some sort of intermittent voltage drop during cold boots. But I haven't noticed any difference when I had only 3 attached. I guess I'll check again.

I opened a thread on Gigabyte's semi-official support forum about this:

http://forums.tweaktown.com/f6...nt-hold-400-fsb-26781/

 

[toreses]

I have a MSI NEO2-FR, and it double boot before
but since I changed a set of memory and change the FSB ratio to 1:2 ( I di not oc)
it hasn't double boot any more...

 

[Bigs11]

I was pretty convinced it was my P31 chipset just not being up to it.
My next board was therefore going to be the P35 (in Gigabyte).
But after seeing similar problems on the P35, I don't know.

I've just been throught the morning exercise of trying to hold 3.6Ghz on cold reboot.
I ended up with 1.47v Vcore, FSB +.2, MCH +.2..... still wouldn't hold.
So I dropped FSB back to 438x8 (3.5Ghz), and it fired up straight away.
So, I'll try that for a few days, and if it stays stable on cold reboot, I'll start dropping the Vcore back, then the MCH.
I really think the FSB needs that extra, tho.

 

[toadeater]

I still haven't found an answer. Whatever is happening on my board, it's tied to the FSB somehow. 364x9.5 (3.5GHz) will boot every time, but 400x8.5 (3.4GHz) will not, even 365 won't. At the same time, I can run the RAM at 5:6 (~425 FSB) and it will still boot. So it doesn't seem to be the RAM or the CPU AFAIK?

I think maybe it just comes down to the BIOSes on these boards having some kind of bug or incompatibility? Abit had one with the IP-35e and they fixed it, but there hasn't been any word from Gigabyte about this yet. I *know* this is happening to a lot of people because there are numerous threads about it. Either cold boot problems, or problems coming out of S3 standby. It's been happening on some Asus boards too.

 

[OBWan]

I had a similar problem with my P35-DS4 and I found the following thread that temporary fixed the problem: http://forums.tweaktown.com/f6...h-voltage-0-2-a-25469/

The tricky solution is to set PCI-e freq. at 101MHz, then the double boot (along with all MIT settings eradicated) problem is gone.

HOWEVER, people in that thread also mentioned that if the MCH is set at > +.2v then the PCI-e link width will scale down to x1. Unfortunately, I could also confirm that is true. I tried every increment step of the MCH value all the way up to the top +.375v and nothing could change that horrible bandwidth fall back.

So the moral: if you want graphics performance, you can't OC your CPU too high. In my case, I need about +0.2v vMCH to stabilize my Q6600 at 3.4GHz. If I want to go higher I will have to move the video card to the x4 slot.

I believe that this is a BIOS issue in that something got messed up. Hopefully Gigabyte engineers will come up with a solution or work-around to fix it soon. Though, my hope is not high as the thread above was dated in October last year i.e. 7 months ago and the problem still exists.

 

[toadeater]

Originally posted by: OBWan
I had a similar problem with my P35-DS4 and I found the following thread that temporary fixed the problem: http://forums.tweaktown.com/f6...h-voltage-0-2-a-25469/

The tricky solution is to set PCI-e freq. at 101MHz, then the double boot (along with all MIT settings eradicated) problem is gone.

HOWEVER, people in that thread also mentioned that if the MCH is set at > +.2v then the PCI-e link width will scale down to x1. Unfortunately, I could also confirm that is true. I tried every increment step of the MCH value all the way up to the top +.375v and nothing could change that horrible bandwidth fall back.

So the moral: if you want graphics performance, you can't OC your CPU too high. In my case, I need about +0.2v vMCH to stabilize my Q6600 at 3.4GHz. If I want to go higher I will have to move the video card to the x4 slot.

I believe that this is a BIOS issue in that something got messed up. Hopefully Gigabyte engineers will come up with a solution or work-around to fix it soon. Though, my hope is not high as the thread above was dated in October last year i.e. 7 months ago and the problem still exists.

Yeah, I tried raising the PCIe speed to 102, it didn't make a difference. I think the MCH needs more voltage for quad-cores. It might have something to do with the GTL timings (google about that).

3.4GHz isn't bad for a Q6600 though, you're using 450 FSB for that, right? I can't set my FSB higher than 364MHz no matter what CPU speed, RAM timings, of voltage I use, otherwise it reverts to stock FSB after a cold boot 99% of the time. It has to be a BIOS bug because it only happens on cold boots.

 

[OBWan]

The lack of northbridge chipset controls across the boards of this Gigabyte series is, well, quite a shortcoming. Since every single P35 chipset could be different (i.e. manufacturing variations), even if ever so slightly, without offering several fine tuning options for the chipset (e.g. GTL et al) from the manufacturer we are buying the boards out of blind luck - if the manufacturer's preset values hit the right spot we get a super sweet overclocker, otherwise we end up with a mediocre board. For now, I am looking at DFI's options...

My P35-DS4 tops at 460FSB (memtest stable) for Q6600 (confirmed with 3 samples), unlike some others I've seen around 475 - 485. That 3.4GHz setting is done by 425x8 with a pair of Ballistix PC6400. I've tried higher but then it needs more vMCH (which infallibly scales down the PCI-e's link width). Other people of the same board can get up to 3.6GHz with their vMCH left at normal, which makes me jealous. I guess we have to accept what we got unless we can find someone who knows how to hack the BIOS for us.

 

[cyberfish]

My GA-EP35-DS3L "double boot"s too, but everything sticks, and I have heard that it is normal. I am running at 500 FSB (with lowered multiplier on my E6300, and DDR2-800 RAM overclocked at 1:1). Would not POST at 510. No problems so far.

 

[Drsignguy]

Originally posted by: cyberfish
My GA-EP35-DS3L "double boot"s too, but everything sticks, and I have heard that it is normal. I am running at 500 FSB (with lowered multiplier on my E6300, and DDR2-800 RAM overclocked at 1:1). Would not POST at 510. No problems so far.

That it should as when you make a change to your settings it shuts down after you save and then reboots. That should be the norm. The "stick" part is what your hoping for.

 

[cyberfish]

I meant my motherboard would shutdown, reboot (fans spin up, but nothing on the screen), shutdown again after ~3 seconds, and then reboot again after another 3 seconds (this time with POST and everything normal).

 

[Drsignguy]

Originally posted by: cyberfish
I meant my motherboard would shutdown, reboot (fans spin up, but nothing on the screen), shutdown again after ~3 seconds, and then reboot again after another 3 seconds (this time with POST and everything normal).

Repeat:



That it should as when you make a change to your settings it shuts down after you save and then reboots. That should be the norm. The "stick" part is what your hoping for.

[/quote]

 

[cyberfish]

ah, I thought you counted the reboot when exiting BIOS. Nevermind then.