The Ultimate Dream Computer

Ummm, what integrated graphics?

 

Some is 32MB, others can use up to 256MB, but remember integrated graphics has no memory of its own, it uses your system RAM.

 

Why are we talking about integrate graphics?

 

No idea.

The lol thing about a 15Ghz P4 is even a lowly Q9300 at stock speed is faster in the right environment!

 

Well it seems my 3.8Ghz X3220 is not stable unless I use AUTO for the NB. Thats gay.

 

Is that because auto uses a huge voltage?

 

I dont know. Oh well. I'll try later. I'm just trying to get it stable at the moment.

If I cant I might just stick to 3.7Ghz. That's respectable.

 

3.7 is certainly admirable. I was thinking about it, 3.6Ghz is probably as high as I'd ever get a Q6600 G0 with a voltage I'm happy with (1.45 or less - not just for electronmigration but also cooling reasons), and if I got a Q9450 I wouldn't want to raise the voltage at all for fear of electronmigration, so that would probably only make 3.3Ghz, which by my maths makes it almost the same speed. No need to spend the extra...

 

oh wow. dude you got the Q9450 to 3.7 stable...on a P5K PRO. well done. thats brilliant.

 

lol it's a Q6600 xeon. I actually got it stable on 3.8 in the end but the volts were too high.

3.8Ghz on Q6600= Pretty damn impressive in it's own right haha.

 

Well that's it, a realistic overclock is something you can run every day, and if you're not happy with the voltage, it isn't realistic, which is why killerwatt at fragsoc got told off for his 3.7Ghz Q9450 - 1.5V, and on a 45nm chip no less... :O

 

I got 3.6 on 24/7 OC on my Q6600 easily (on Air, 790i). It's all in the GTLVREF Lanes. I had to set Lane 0 & 1 to ~100mV to get it stable in OCCT, and Lanes 2, 3 to about 40mV. I think I could get 3.8 if i tried harder,which I'll have to do during the school year.

If it has been stable on full load for 6-8 hours (assuming thats a good buffer). It doesn't matter what the volts are if its stable, if its stable it means no errors are generated when you run a cpu intensive program. The only thing high volts do is shorten the life (or make it more unstable, which you checked for)...so instead of getting 15 years out of a processor it might only be 3 or 4....which by then, if your an enthusiast, you should have a newer processor.

 

Actually no, with a 45nm Quad, it's months, sometimes only weeks. This only occurs if the CPU is hot of course, and 65nm chips are far less vulnerable, but get a 45nm Quad to 1.5V, unless it's running at ambient or below, it won't last very long at all.

 

The absolute Maximum Vcore for the 45nm is 1.45V, im not surprised it would only last a little bit. But if your under that you should be okay. It does only make sense that a smaller amount of metal will deform faster than a larger amount of metal under heat?

But what I was trying to get at, is that if it's stable for a long time...like a day or so, then you should be good for awhile, because chances are your processor doesn't run at max stress all time...unless you game/run folding at home/other application for more than a day, or close too...especially with newer/faster cpu's, you shouldn't be at max stress for than a day. If you are, you are going to have to shoot for lower overclocks than the regular overclocker.

 

It doesn't matter. Even if it's not full load you're still running the same amount of voltage through the CPU.

 

Indeed, but most overclockers here laugh in the face of 'chances are' - I'm not one of them, but some here are very stringent about what they consider stable.

 

You mean voltage "across" the cpu, but anyways...so you're saying you run your CPU at max stress for more than 24 hours regularly???
For the avg user...a 6 hour stress test is good enough, but most people recommend you run a cpu stress test for a solid day. It does always depend on the user and how often there cpu is under stress, but if your running at full load for more than 24 hours...your CPU isn't gonna last that long anyways, so whats the point of OC'ing? Thats my view, , and it doesn't seem naive to me at all.

 

Through, across, same thing.

Look, even if you're NOT RUNNING at load, you're still running the same volts. Meaning the same chance of elecromigration that will destroy your CPU.

I don't keep the same CPU for more than 3-4 months, and I still don't risk a Q6600 above 1.5v.

 

I have my 2.66GHz Q9450 at 1.3v which is 3.4GHz, thats all I'm going to push it. 3.6GHz shouldn't be a problem BUT why chance it. keeping an eye on your temps will tell you alot

 

See how much you can go without raising the volts. Thats the key. Don't raise the volts too much and keep your temps under check.