YiffyGriffy

I believe it just runs them for xx number of minutes, by which the default is 15 minutes.

__________________
Went almost two years without any major system overhauls, so I'm treating myself

Melraidin

Yep, the default is 15 minutes, but that's not the number of minutes to run each test. That's the number of minutes to run each different size of FFT (8K, 16K, etc.). So still not sure why it is that the tests run so much faster at lower CPU speeds...

Wondering if perhaps the programmer of Prime95 could explain.... I've had a very quick look through the source and haven't noticed anything peculiar, but haven't had a chance to read it in detail yet.

YiffyGriffy

You should email him/them and find out, Still haven't had a chance to try it myself (still burnin in 2640Mhz oddly enough, I'm going backwards), but I might later tonight.

__________________
Went almost two years without any major system overhauls, so I'm treating myself

Melraidin

So you're falling backwards as well... What the hell's with these chips...

I had 2695 MHz, now I'm fighting for anything over 2640. This is bloody ridiculous!

BTW, also been noticing some strange things with voltage regulation to the CPU. Pumped this to 133% of 1.55 V, so that should've been about 2.06 V. Barely made it over 2 V according to SmartGuardian. Then dropped back to 126% of 1.55 V and all of a sudden it shot over 2.0 V in SmartGuardian. Very strange. Now I'm at 113% of 1.55 (I think) and SmartGuardian's showing 1.71 V... Better, but can't get stable at 2695 MHz still, even though my temps're now -5 to -6 C!

YiffyGriffy

Yeah I'm not real up on that VID Special crap, I did 1.55 and 113% which should've been 1.75, but gaves me 1.82, too much. I'm running 1.475 and 110% which gives me about 1.65 for 2640Mhz. I'm running lower cause I'm trying to figure out why it won't run 2.7 (2695). It passes Prime, but still crashes periodically so I think it's the RAM so I loosened some of the other timings a tad and making my way back up.

__________________
Went almost two years without any major system overhauls, so I'm treating myself

Melraidin

K... this has definitly gotta have something to do with the chip I think, either that or the motherboard. Just tried again at default speeds (800MHz) and saw the Prime95 tests blazing past. Roughly 1 - 2 seconds apiece. Now tried raising the CPU multiplier from auto (4x) to 5x. Now the tests are taking far longer than at 4x, haven't completed one yet but I'm guessing it'll also take longer than at 2600 MHz.

There's gotta be some sort of explanation for this somewhere....

YiffyGriffy

Now you're talking about the torture test, right? Not the Benchmark thing?

__________________
Went almost two years without any major system overhauls, so I'm treating myself

Melraidin

Yep, torture tests. Specifically the small FFT test that starts at 8K FFT length.

Right now I'm totally stable at 2640 MHz with 1.76 V on the CPU, yet still haven't figured out this oddity with Prime95.

YiffyGriffy

Well, seeing as how this is the longest I've run Prime95 in my life ~3 hours, I think I'll let it run the rest of the night and check on that stuff tomorrow. 2695 is pretty much a no go, more voltage than I really want to run for 24/7, so I'll take 2640 at 1.65v.

__________________
Went almost two years without any major system overhauls, so I'm treating myself

Melraidin

Well hot damn! Liking the response times on those guys coding Prime 95, got back really quickly! Here's the reply I got when I asked about why Prime 95 was running so much faster at the slower CPU speeds:

From George at MIT:

So I guess that explains that little oddity. Now to figure out if I'm ever going to get this CPU over 2.7 GHz......