Cold boot problem

[thebiz]

I'm getting some strange results OC'ing my 2600+ mobile. If I boot at stock speed and then restart through windows (or the reset button) I can OC it to about 220x11.5 and be prime95 stable. If I shut everything off at that point and try a cold boot, it won't post, and I have to reset the cmos via a jumper.

Details:

2600+ M
DFI Lanparty Ultra 2 B
1x512 OCB 3500EB
Thermalright SP-97
Vantec Tornado
Antec True480
HTI Radeon 9600XT
VGA Silencer
WD Raptor 10K 36
Maxtor 7200 200GB

At 220x11.5 its at 1.85 vcore, 2.9 ddr voltage, and 1.8 on the chipset voltage with loose memory timings. The AGP/PCI busses are locked at stock speeds. It runs cool (low 40's under load) and is prime95 stable, but won't cold boot. The voltage on all the rails stays stable under load, so I don't think it's the psu.

There are enough variables that could be at work here that I'm not really sure where to begin. Any ideas?

 

[ectx]

What is the highest speed at which you can cold-boot? How do you tell 220x11.5 is really stable? In the ancient time I know of a few mb's that had cold boot problems (some have solutions and some don't). I don't know much about DFI Lanparty so I could not tell.

Last resort (if it is stable and could not find any problem for cold boot): use software controlled overclocking and set it to oc everything you start windows.

 

[thebiz]

The highest cold boot speed is about 205x10.5, although this seems a bit variable. I'm judging 220x11.5 to be stable because prime95 runs overnight without any errors.

I assume that using software controlled overclocking is inherently not as good as hardware, which is why it would be a last resort?

 

[ectx]

Originally posted by: thebiz
The highest cold boot speed is about 205x10.5, although this seems a bit variable. I'm judging 220x11.5 to be stable because prime95 runs overnight without any errors.

I assume that using software controlled overclocking is inherently not as good as hardware, which is why it would be a last resort?

I occasionally had probelm w/ software based oc when I adjusted the fsb by too much. Other than that I cannot say software oc is worse that bios oc. What I meant is I would try to debug the cold boot problem more before being convinced that there is no other problems.

 

[Shimmishim]

the cold boot problem is related to the bios and not the overclock.

go visit amdforums or pcperspectives or whatever it's called now...

there are plenty of topics on this cold boot issue with the dfi ultra infinity / lan party

i used to have cold boot issues also until i switched over to the 1/21/2004 bios and i havent had any boot issues since...

 

[superHARD]

I bet the voltage is applied after it posts...therefore the cpu doesnt have what it needs to start the first time.

 

[Jgtdragon]

Does this happen with stock speed?

 

[thebiz]

i used to have cold boot issues also until i switched over to the 1/21/2004 bios and i havent had any boot issues since...

Excellent, thanks for the tip. I flashed the bios and it seems to have alleviated the problem.

 

[DaNorthface]

you could just turn on the computer - if it doesn't boot, press reset and it should work that time around. Also you could hard mod your voltages on your cpu if all else fails. My friend's shuttle mobo had this problem, and i read/tried alot of things to fix it. Some of the things i read that worked for others besides the ones i mentioned above is: inserting the memory in 1 and 3 slots, taking out the memory and putting it back in to scrap off any oxides, reseting the bios to safe settings - and then changing it back to whatever you want, and trying different memory..

 

[Shimmishim]

Originally posted by: thebiz

i used to have cold boot issues also until i switched over to the 1/21/2004 bios and i havent had any boot issues since...

Excellent, thanks for the tip. I flashed the bios and it seems to have alleviated the problem.

good stuff!

glad i could help