Overclocking PCs

From w.mfi
Jump to: navigation, search
HWiNFO and AIDA64 stats on my CPU while running the x264 stress text

I put together a custom PC and overclocked it from 3.4 to 4.4 Ghz, with the great help of folks at Overclock.net, especially Darkwizzie's Haswell guide. I posted my specs there and also went on to publish some stats on the parameters overclockers chose.

Ultimately I documented 185 tests and put together a sensible man's guide to overclocking.

A vocal minority of the overclocking community thinks that the only way one should test their rig is to use the most torturous test they can devise, for long periods of time. But I say they're pointlessly crippling their build; a good choice is much simpler and easier than that. Just use the most strenuous real-world example you personally might actually encounter (like video encoding), find the point where it's stable overnight, then give your CPU one final stability bump (+0.04 VID or so). Slightly backing VID off from the edge of instability is a quick, easy way to cut straight through all the uncertainty and go on with your life.

They scoffed, saying the only way to test is with the toughest challenge imaginable (which greatly overheats their CPU, which means they have to scale back their clock rate, the antithesis of optimal reliable overclocking). In reply to one saying only the worst test could make it completely stable, I said

Good enough; that makes sense just fine. I am not saying it's wrong.
I'm saying that there's a better way.
One is the absolute approach, the other is relative to the work expected of the CPU.
Here's the absolute argument as I see it; please correct me if wrong:
Realistically, you can't test an OC forever to make sure it will never fail.
So instead, you make it pass a more stressful test. Indeed, the most stressful test you can find.
If it can pass the worst stress, it should be ok for everything else.
Good enough. It makes sense just fine. None of these arguments are rocket science.
Stats from overclocked Haswell CPUs at Overclock.Net
But there is another way.
You can chart the CPU's stability versus volts, for work you will actually do. As part of making this line, find a voltage stable for a long time (e.g. overnight).
Then give the voltage one more nice bump up the stability line you've drawn.

So, while you can't test forever,
You don't have to.
You don't have to cripple your actual work, to be absolutely theoretically sure it can handle everything.
Of course, if you WANT to be absolutely theoretically sure, then go for it. It's your choice.

A number of people liked my thread, and it had a life of its own long after began it, then moved on.

To me, it still feels like a real breakthrough in understanding the practical implications of what you're trying to achieve, and how to do it. I'm kind of proud of myself, to think I wrote up a better way to overclock than all the gearheads in the world had. I'm not saying I'm the only one to ever think of it; how could I know that? Plus I stood on the shoulders of the x264 test. I'm just saying that I'm the guy who wrote up a neat little guide that ties everything together conceptually and practically.

It's been three years since I posted that thread. I since found my PC to lock up on one or two games a year. The solution was super simple: Knock the overclock back to 4.2 instead of 4.4, until I finished that game. Easy peasy.

Best wishes to you if you're overclocking!