Crash after cold start, then runs forever on 2nd start

[hkancyr]

I have the fantastic Badaxe2 with a QX6700 4 Gig's of twinx 6400C4 and 2 RAID 0's and 1 RAID 1 and 2 single drive, 650W PSU and a 8800GTX 768.
After a month of trying to install VISTA64 including going to VISTA32 for a short time, I have VISTA64 running beautifully, it does everything except run my TiVo server. I am running all stock speeds and voltages, except memory which is at 2.0v.
The problem is that I ALWAYS have to start it twice. On the first start it runs fine and all for about 10-20 minutes and then locks up and makes a click noise on the speakers as it does. Then I restart and it runs forever like there was no problem at all. I can play Farcry for a week and it won't even burp.
I have tried tweaking voltages in a very small way but nothing seems to help.
I just have a feeling that there is a setting in the BIOS that will set me free from this annoying problem, but I can't figure it out and I haven't seen this mentioned in the forums anywhere either.

P.S. I just found this forum as a result of my searches, I wish I knew about it before, seems to be more XBX2 people here.

[hkancyr]

I may have this thing beat. I pushed the cpu voltage to the max mentioned on the box 1.35v. My cold start was an hour and a half ago and it hasn't misbehaved yet.
Yesterday I found my memory timings were low

this is where it locked up! I can hardly believe it..damn. Just as I was typing this.

Oh well, I must be on the right track, so now I just kicked up the voltage to 1.3625, what must be the danger zone, in the eyes of Intel anyway. Perhaps my PSU is not supplying the correct voltage or the mobo is somehow sucking some of it away for it's own evil purposes. Speedfan is telling me that I have 1.17 at Vcore. The temps are running 30-35c at idle, internal temp is 42c. 41-52c playing Halo2.
Getting back to the memory, I was running it at 5-5-5-15 which I saw on the Corsair sight a couple months ago and I just checked again and they say run it at 4-4-4-12 and 2.1v. So, OK, I did that too. So now I'm wondering if I should play with the MCH/ICH voltage too.

I have to shut it off for an hour or so to see how the next cold start will go.

[Apoptosis]

strange stuff... do you have a volt meter?

[hkancyr]

Yeah I have one. I also have a cheapo PSU checker the Coolmax PS-124, still in the wrapper.
Does Speedfan indicate this lower voltage for some throttling reason or another? It shouldn't be stepped down when Halo2 is running though, should it? I just popped Speedfan on, it is indicating Vcore 1.32v now. Core temps running 36-42c at idle.
Anyhow I am in the waiting for disaster mode right now, I started it around 15 minutes ago. If all goes as expected it should run for longer than 2 hours this time. What seems stranger is that it seems to crash so often in the middle of a post that I am typing.
I'll check back later.

[hkancyr]

The PSU checks out fine according to my little tester.
I have adjusted my Vcore voltage lower now to 1.3v. Higher voltages didn't seem to help much.
I adjusted the FSB voltage down as well to 1.2v. Seems to boot a bit slower but it boots and runs OK.
The problem still exists however.
Oddly, after these voltage changes there seems to be no change in the Speedfan readings. It still says Vcore is between 1.14v and 1.23v depending on whats running.
I continue to be baffled.
I will continue chasing this thing too.

[hkancyr]

I tried over clocking to a 12x multiplier, upped the core voltage to 1.35 and no change, locked up 10 min after cold start. Then of course ran super fine after that.
I'm kinda running out of BIOS options.

[hkancyr]

I think this problem is fixed.
One of the RAID0 hdd's was screwy. I recently reinstalled using the other two SATA (intel) ports, the south 2, and 2 Maxtor 500's, no problems for 3 days and before it would crash 30 or more minutes after a cold start.
I don't know if it was the ports or the hdd's, but one of them was the problem. I have not plugged in the problem drive since then either, but have since reformatted the "good" Hitachi 160 and used a north port for it.

[Nein]

I've a suggestion you may or may not found useful, do you want to hear it since there's no other?

[hkancyr]

Go right ahead, but it seems to be fixed.

[hkancyr]

My problem returned after updating BIOS to 2809. I returned to 2802 and the problem went away again. Looks like I'll be staying with 2802 unless something remarkable is fixed with a newer BIOS.

[hkancyr]

Disregard previous post. It just did it again with the old BIOS. I just got a 950 watt PSU the other day as a backup. I guess I'll drop that in and see how it goes.
Sorry for the bad info.

[hkancyr]

I believe the situation has resolved itself. All the tweaking I did had very little impact evidently, because now all settings are stock except mem voltage. It must have picked up an update to something or one of the SP1 betas did it. I am now running the real SP1 and it's still running fine and dandy. Anyhow, it's been running crash free for about a month now.
Thanks everyone for your help.

[DMB2000uk]

Glad we could help >_<

Sorry there weren't more helpful suggestions!

Dan