Motherboard "forgets" ONLY overclock settings after power off

[Al_aric]

I can get a stable 4.0 ghz OC (14-18 hrs Prime95 Blend setting) on my e8400 with the following settings (Gigabyte DS3L mobo):

CPU Bios VCC = 1.300 ----> CPU-Z voltage reports 1.248 under load, 1.264 at idle
PC1066 DDR2 @ 2.14v BIOS -----> 2.1v through Everest (or like SW monitoring utility). OZC recommends 2.1-2.2, so I went with 2.1 if proven stable.
MCH= 1.2v in BIOS. Can't find a utility to report voltage in Vista.

Note: I don't push my MCH higher than 1.2 because intel spec says don't go above 1.21 AND I don't have physical space to zip-tie on a fan. Stability seems unaffected when running FSB at only 445 mhz with 1:1 divider (890 mhz RAM clock).

So if all this is stable under stress testing, why does my BIOS reboot the PC 4-6 times before deciding to reset to defaults IF and only IF the system has been powered down for a couple hours? It doesn't happen when the system has been off for only a few minutes or even an hour. Also, the MOBO is brand new so it isn't the battery (if the board even has one...) and also, when I don't overclock and stick to auto-detected settings, this problem doesn't happen. It has to do with overclocking. Mild overclock (from 3.0 to 3.6 ghz by bumping FSB from 333 to 400 mhz) DOES NOT trigger this problem, either. It seems to be only after hitting 4.0 ghz.

Help. Please.

 

[RJR]

Just curious, where did you come up with 1.21v max NB voltage?

Here's a couple of things to read:

http://www.ocforums.com/showthread.php?t=546768
http://www.hardforum.com/showthread.php?t=1193054
http://forums.whirlpool.net.au/forum-replies-archive.cfm/1059673.html
http://www.tomshardware.com/forum/256936-30-gigabyte-cold-boot-issue
http://forums.overclockers.com.au/showthread.php?t=668920

 

[Al_aric]

The 1.21 max voltage comes directly from the Intel spec sheets on the website under P45 technical documents. Now, of course, these are the same documents that say you can't run a .45nm Core 2 Duo at more than 1.3625v and some of us do (not me, I make 4 ghz at 1.25v)... but considering I don't have the physical room to actively cool the Northbridge, I'm going to stay within the limits.

So far, 1.21v has been feeding my 445mhz FSB requirements just fine, aside from this damn boot problem.

 

[chunkymonster]

It could be a psu issue, either not delivering enough watts or enough amps on the 12v rails.

Verify that the BIOS jumper is set to retain BIOS settings changes and not in the RESET CMOS position. By having the BIOS jumper in the RESET CMOS position, the BIOS will "forget" any BIOS changes when the machine reboots.

If the machine reboots when overclocked to a certain level, it could be heat issue as well.

List complete system specs for answers that are more than guesses.

 

[RJR]

same documents that say you can't run a .45nm Core 2 Duo at more than 1.3625v

The 1.21v is the absolute Max, the max listed is 1.155v

Just like the 1.3625v is the max and 1.45v is the absolute Max listed by Intel for the 45nm's.

I'm running an older P35 at 1.4v and Intel states 1.3v Max but with 8 Gb's Ram that what it wants and haven't had any problems yet.