Testing overclock, help please.

Armadillo

Armadillo

Armadillo

Soldato
Joined
29 Feb 2004
Posts
3,856
Got my x4 955 set to 3.8Ghz. Just wondering what is the best way to test it and make sure it's 100% stable? Looking around, I see a lot of differing opinions on what needs to be done and for how long. At the moment it's set to 3.8 and the voltage is done via the offset so shows 1.41ish in bios and 1.44V max under load from linx.

Goes through a 20 run of Linx at 20000 problem size fine with these settings, no errors, 55C max. Some posts around suggest this is fine and all that needs doing. Played some hours of bf3 on it as well and no issues.

So fine? other tests to run? Adjustment was made using the multiplyer only, so northbridge and memory are still stock, so don't think they need testing.

Thanks :).
 
it seems ok so its upto you.
you can always try an overnight run of Prime95.
just do the blend test, and an overnight run will be enough imo.

i often just use Intel burn test (nearly same as Linx) with 30 passes, if it passes, i then do an overnight run on Prime95, if that passes as well, then its stable.
 
Spenty said:
I find the best way to test an overclock is to use your system as you would normally.
Click to expand...

As Spenty says but I'd do a quick intelburntest or prime95 for at least an hour if you want to be more sure. Otherwise you might run a game which works fine but that's only because of a GPU bottleneck and then crashes on a CPU intensive game.

If you do want to make 100% sure especially if you're decoding, rendering or what not you should try a length prime95 or longer intelburntests like in the 4+ hour range.
 
Meh, one hour in to prime, come back to check on it and windows has recovered from fatal error :(. Bluescreen view shows same crash as what I had when I tried 4.0Ghz at 1.44 and linx crashed it straight away.

So I assume nudge the offest voltage up a little and try again? Like I said in first post, it's mulitplier only at the moment, northbridge at stock and memory at stock. I see posts saying nb might need more voltage, but I think this is only for when this is overclocked as well? Not when it's left at stock?
 
Last edited:
1.452V and still not having it, crashes out and reboots before even 1 hour has passed :(.

Suggetions? Nudge it up once more and try again?

Really would like to get it 3.8 stable. Is something else failing, rather than cpu (everything else is stock) or do I just have a crap one that isn't going to do it (That's normally my luck).

Thanks.
 
Last edited:
I'm not familiar with your chip I'm afraid, but for reference I recently got my Q9550 stable at 3.8GHz. I almost had it stable - Intel Burn test stable, P95 large and small stable, but blend would fail. What fixed it for me was upping the PLL voltage a notch, but I would highly recommend you read up on it regarding your chip, for me I saw numerous warnings about going to 1.7V and above on 45nm Core 2 Quads (I bumped mine from 1.5 to 1.6 to finally get stable).
 
From looking online, equivelent for amd is vdda voltage, which I have not touched, just left on auto. No idea what it can safely go to, so I'll have to look into that, I was hoping to just get a quick and easy overclock. Increase multiplier, increase vcore and hope it worked.

I've only touched the cpu Vcore, which can safely go to 1.5 according to Amd. I did try 1.46, after 1.45 failed and 1.46 primed for 6 hours before crashing.
 
You must log in or register to reply here.
Back
Top Bottom