Tricky Overclocks

[ziddey]

Anyone ever experienced overclocks that can be stable for 24+ hours if you don't touch it, but otherwise, it's anybody's game?

When I clock to 8x450=3600 with my e6400 with 1.475v (no drop/droop) and ram 1:1 4-4-4-10 2.29v d9gmh on a p5b vanilla (did same settings on ds3 with identical results), I can prime small/blend/large for almost as long as I want if I don't touch the computer while it's going.

If I use the computer while it's going, there's a good chance there'll be a bsod (usually machine check) at some point. Today I got to the 14hour mark in orthos (usually I've found crashing at 8 and then 14 and then around 24). While I was loading a website, it just up and died on me.

When I'm at 8x400 with 1.325, I'm rock solid forever and a day no matter what I throw at it.

Temps could be a little better, and I'll improve them soon. But currently, I'm loading in the mid 60's. It used to be in the mid-upper 50's before when I used a different paste and room temp was lower.

8x400=3200 1.325v, I get load around 50 or so. All temps are coretemp.

I haven't measured vfsb or vmch on the p5b yet, so I'm not sure if they're setting properly. I did verify on the ds3 when I had same symptoms.

I've got a 120 spaced above the ram blowing down, and it's keeping them real cool.

psu is enhance 500w

Anyone have similar situation? Or at least an ability to be able to detect this error quicker, instead of having to orthos or otherwise use the computer for a day or two before it randomly crashes? Obviously, occt says it's stable.

 

[Arcygenical]

Make sense...

When you're using the computer you stress the IO system as well, something that Orthos doesn't do.

 

[ziddey]

Make sense...

When you're using the computer you stress the IO system as well, something that Orthos doesn't do.

yeah. I mean it seems like I've found the cpu to be "stable". More voltage doesn't seem to help. I've also tried 2 pairs of d9gmh's and at one point have determined validity in each.

I've used 3 different motherboards (1 ds3 c1, 1 ds3 c2, 1 p5b vanilla c2). I'm still hoping it's just three unlucky motherboards and hope that when the dfi infinity finally comes out, it'll be promising.

I don't really feel like pulling out the ds3 anymore since I'm pretty sure it was responsible for nuking my ram. I never used insane voltages (2.2v on the ds3s) but I never had a dmm on fulltime to track if there were spikes or otherwise unclean power.

Northbridge cooling has been more than adequate. I can't imagine the southbridge getting too hot, but maybe I'll rest a fan on top of that and try again. I haven't probed voltages of vmch/vfsb/vsouthbridge yet, but vcore and vdimm seem to be stable.

Bit by bit, right? I've already got a better psu to be able to rule it out (although it does make a squealing noise, but I'm thinking (hoping) it's just loose inductors and not something that's affecting anything).

The p5b has been a little disappointing so far since it can't go much after 450 without being incredibly instable. And even at 450 (and sometimes 400, although higher vmch seems to cure it), it'll often not be able to post after a warm reboot (even though it still cold boots). A quick reset hit fixes it though

 

[Arcygenical]

I'm suprised you got a P5B-vanilla past 420FSB...

I use the -e variants to get into the 475s

 

[orion23]

I cannot be orthos stable not even for 10 seconds, however, nothing I throw @ my computer makes it crash!

Fvck Orthos!

 

[ziddey]

I'm suprised you got a P5B-vanilla past 420FSB...

I use the -e variants to get into the 475s

lol I can't see vanilla being that bad. Personally, my vanilla had way more bios voltage settings than I expected. However, I'm not sure if they all work as they should or not. I'll verify it with dmm in a few days. That said, I guess ds3 spoiled me when it posted at 400fsb right when I got the system put together.