PDA

View Full Version : Overclocking Question


Bags
2012-12-02, 07:16 PM
Hey, now that I needed to worry about it (PS2 release) I got back into OCing.

I started at 4.3 ghz and it worked fine. Ran prime for 20 minutes and temps peaked at 62. Given the low value I decided to forgo a longer test at that speed and went up to 4.5

However, at 4.5 windows locks up at the logo.

I dialed is down to 4.4 ghz, and it booted fine and currently is maxing at 65C on prime 95.

Any ideas why 4.5 ghz won't boot?

the guide I followed: http://forums.overclockers.co.uk/showthread.php?t=18340310

Ailos
2012-12-02, 07:58 PM
Every processor is simply different, and the combination of voltage and temperature you're setting at 4.5 just doesn't cut it. All it takes is one transistor to fail, but that transistor failing 4.5 billion times every second creates a lot of errors.

You can solve this by slowly bumping up your voltage, but given your current temperature situation, that's a tricky proposition.

Also, the silicon's stability degrades over time. If you've been running at that OCd setting at those temperatures for a long time (18+ months), you'll start seeing your voltage requirement slowly creeping up to keep stability at the same speed.

Bags
2012-12-02, 08:22 PM
Nah I OC'd it for a day or two back when I got it but then turned it back to stock since I didn't need it.

Rbstr
2012-12-02, 11:42 PM
65C is nothing from a solid-state perspective. The silicon had to go to 1200C to get the dopants in the right places and like 400C to anneal the metal layers.
You won't break the CPU until around 100C and that's because of packaging failure.

What would likely cause failure is thermal expansion stress from going up to temp and back down a lot, especially in the contacts.

Ailos
2012-12-03, 02:10 PM
Nah I OC'd it for a day or two back when I got it but then turned it back to stock since I didn't need it.

Then if you have the headroom, nudge your voltage up in really tiny increments, as finely as your mono allows. Eventually, you'll reach stability.