• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Simultaneous torture testing with different programs

While I Am jealous of some of the low voltages and temps you guys report, I am happy with this 9 hour run with P95 and IBT. Temps stayed in spec and no errors reported.
finalbrag.jpg
 
lens pirate said:
While I Am jealous of some of the low voltages and temps you guys report, I am happy with this 9 hour run with P95 and IBT. Temps stayed in spec and no errors reported.
Click to expand...

What is your QPI voltage, still at 1.2 volts? Try setting it to 1.4v and see if you can push cpu-v down -- couple of us found we were qpi-v limited, not cpu-v limited. I found that starting at 3.7 GHz my cpu hits a huge vertical wall of voltage to keep going. You might try 3.6 GHz instead, it may save you lots of voltage, power and heat.

It takes a lot of time to figure out what is affecting stability on i7, so many adjustments...
 
I can do linx 200x19 stable with Vcore @ 1.386v and 1.425v for vtt (qpi/dram for the P6Ts), but with both Prime95 running (8 threads) and linx running together? I got a BSOD 124 error telling me I don't have enough voltage for the VTT (which is at 7.2GT/s). So I'm back at square 1. Is there a safe limit to how much VTT voltage you can add?
 
you didn't say what your uncore frequency was. does your board peg it to 3.6GHz for 7.2GT/s? that's pretty damn fast if so. can you lower it?
 
Sorry, my uncore is at 3200, which is the lowest speed I can go with my 1600mhz rated ram. Even at auto it is at 3200 for the uncore.

As for QPI Link speed, same deal again, 3.6ghz in bios is the lowest (7.2GT/s). I would LOVE to be able to run it lower than 7.2 though, as this seems to be what is limiting my oc. But I'm not sure if this is possible at all with 1600mhz memory.

This is all on a P6T Deluxe. I can't be stable at all unless I push up the VTT aka QPI Voltage :(.
 
Last edited:
xorbe said:
What is your QPI voltage, still at 1.2 volts? Try setting it to 1.4v and see if you can push cpu-v down -- couple of us found we were qpi-v limited, not cpu-v limited. I found that starting at 3.7 GHz my cpu hits a huge vertical wall of voltage to keep going. You might try 3.6 GHz instead, it may save you lots of voltage, power and heat.

It takes a lot of time to figure out what is affecting stability on i7, so many adjustments...
Click to expand...

This post was a GEM! As it turns out my QPI voltage which was set to auto was a limting factor!!!!!!!! Woot. I raised my set my QPI voltage to 1.39 and WHAM! all of a sudden
3.8 ghz at a cpu voltage of 1.268 was fine. This lets me run my memory at 1600 hundred and 1.64 volts. Temps stay perfect not going over 74c on the CPU.

Man the I7 processors are amazing and will over clock like crazy but damn... It sure is more complicated than other chipsets of the past.

I need to do some testing at lower voltages... I may be able to go higher. But if not that is OK.:D
 
lens pirate said:
This post was a GEM! As it turns out my QPI voltage which was set to auto was a limting factor!!!!!!!! Woot. I raised my set my QPI voltage to 1.39 and WHAM! all of a sudden
3.8 ghz at a cpu voltage of 1.268 was fine. This lets me run my memory at 1600 hundred and 1.64 volts. Temps stay perfect not going over 74c on the CPU.

Man the I7 processors are amazing and will over clock like crazy but damn... It sure is more complicated than other chipsets of the past.

I need to do some testing at lower voltages... I may be able to go higher. But if not that is OK.:D
Click to expand...

What are you using the test to ensure stability? 3.8GHZ at 1.268 vcore is amazing. I wish my setup could do that.
 
lens pirate said:
LOL I just did one better...check out linx stable 3.8 at 1.256 Vcore. Look at those Air cooled temps. see picture... WOOT!



Load1.jpg
Click to expand...

What a lucky guy you are. Even with 1.4v on the qpi I cannot use anything below 1.36v on vcore otherwise it will bluescreen or error out. Linx + prime is a killer.

Did you tweak your cpu pll, ioh/ich voltages?
 
Nope I tweaked the CPU VCORE, DRAM and QPI. I use a 19 x 200 multiplier. I do not let the CPU idle down, locked at 19.
 
lens pirate said:
Nope I tweaked the CPU VCORE, DRAM and QPI. I use a 19 x 200 multiplier. I do not let the CPU idle down, locked at 19.
Click to expand...

I am jealous. You're quite lucky.

Though in reality anything beyond 3.6ghz will not really net you much extra performance. But then again, no program I will ever use will push temps as high as LinX. Just not possible.
 
hazard_one said:
So why hasn't this method of testing been made the standard?

I'm sure plenty of people have thought about doing it. Even from the beginning.
Click to expand...

Probably because it wasn't necessary... It was much easier to fully stress test dual-core rigs and even C2Q's using a single stress test or a stress test + some routine stuff (movie/music playback and/or 3dMark or w/e), the i7 being so much more powerful and having HT is much harder to really push to the edge. Between QPI and the integrated memory controller there's also a lot more stuff to tweak. Makes perfect sense to me, looks fun too, I don't see why some are calling it madness.

I'd rather mess around w/my system for a week after having built it, stress testing it back and forth to the brink, than to have to deal with some obscure error or highly inopportune crash months down the line, at that time it might not even occur to you that it could be the OC because it's seemed so stable. That new version of OCCT with the GPU memory tester and the PSU-mode testing looks pretty interesting btw, has anyone given them a try? I assume someone has since it came out in Jan when this thread was started, what's the verdict?

DeadSkull said:
Well in my case I only have a Q6600 so I can't run IBT/Prime95 at the same time, but I've dabled in extreme stability. My Q6600 @ 3.897 passed 103 IBT runs stable and later to insure stability I ran two instances of Prime95 for a period of 24 hours during which I also fired up Furmark several times, one time for a period of 3 hours which my PC handled no problem.
Click to expand...

Surely that's not considered extreme now days, is it? I thought it was pretty much common practice to get Prime95 running for 12-24 hours + something like Furmark or 3dmark (or rthdribl before any of that was common) to insure stability. How else are you gonna test all your components at their maximum heat output otherwise? (which a really demanding game may cause at any given time, or a game + w/e background tasks)

For the nay sayers, try any of this on a stock-clocked system, unless it's having thermal issues it will never buckle... The processors probably go thru a heck of a lot more when they're binned and stress tested at Intel/AMD. Do you think cars are tested at a leisurly pace down the neighborhood before they head out the factory? :p People aren't stacking programs on top of programs here for the sake of it, they're just trying to saturate every component and sub-system simultaneously. Something none of the stress tests can really do by itself. If you read thru the thread it makes perfect sense, stuff like Prime 95 just adjust on the fly and takes up w/e else IBT or Linpack leaves free.

P.S. I do have one critical comment regarding some of the tests done here though... You gotta be ballsy to stress test on a beta OS. ;) Granted, you're stress testing the hardware, not the OS, but it's still one more odd variable thrown into the mix, heh; 'least driver conflicts are unlikely going from Vista to W7.
 
Last edited:
actually the manufs know where the speed paths are, so they know the exact instructions to test ... but like hell they'd ever reveal the speed paths or the tester programs. It costs money to test each chip, the faster the better. It's not an issue unless they get aggressive. Like old Cyrix CPUs, or 1GHz P3s, or today's "super clocked" video card GPUs...
 
You must log in or register to reply here.
Back
Top