I7 / P6T - Using SKEW and SPREAD to increase/stabilize OC THREAD

[286SX]

I have been able to use relatively low voltages to get a stable 4.0Ghz OC (=11+ hours prime95 blend stable @ 1.3125v on Cooler Master V8, closed Antec Nine Hundred case, 71 degree ambient). As I was able to achieve stable 3.6Ghz @ 1.14v, I suspect that 4.1+ is achievable without pushing 1.4v. [and 4.0 may be stable with less, but the lowest I have tried is 1.3125v. See here].

However, in my first few attempts I could not get a prime95 stable 4.1Ghz or 4.2Ghz OC using voltages up to 1.35v. At 1.35 (and 1.325) I can get Windows to boot and run most applications, but prime95 reboots in less than 1 hour. Clearly, one reason could be because of low voltage.

Thus, the low-ish voltage issue gets us to the meat of the matter. Has anyone played with the advanced OC tuning features in the P6T bios [including skew and spread] to attempt to reach more stable and/or higher OCs [with or without lower voltages]?

I will post my results here, and would appreciate others doing the same.

Cheers,

-td

[GenTarkin]

I have been able to use relatively low voltages to get a stable 4.0Ghz OC (=11+ hours prime95 blend stable @ 1.3125v on Cooler Master V8, closed Antec Nine Hundred case, 71 degree ambient). As I was able to achieve stable 3.6Ghz @ 1.14v, I suspect that 4.1+ is achievable without pushing 1.4v. [and 4.0 may be stable with less, but the lowest I have tried is 1.3125v. See here].

However, in my first few attempts I could not get a prime95 stable 4.1Ghz or 4.2Ghz OC using voltages up to 1.35v. At 1.35 (and 1.325) I can get Windows to boot and run most applications, but prime95 reboots in less than 1 hour. Clearly, one reason could be because of low voltage.

Thus, the low-ish voltage issue gets us to the meat of the matter. Has anyone played with the advanced OC tuning features in the P6T bios [including skew and spread] to attempt to reach more stable and/or higher OCs [with or without lower voltages]?

I will post my results here, and would appreciate others doing the same.

Cheers,

-td

I have a gigabyte EX58 board and it has some skew options as well....I was wondering the same thing.

[STi489]

Spread disabled or not doesen't seem to make much of a diff here..I have not messed with SKEW.

[JWMc]

Someone please post some information about what the Skew settings do and when you might need to change them and in what increments, upper/lower bounds, etc.

I would also like to learn if any of the PLL voltages are actually useful on the i7 platform. I've seen people post that they are using core and/or QPI PLL voltage changes in some cases but they didn't post any info that would tell you that they scientifically determined that they actually needed the PLL changes to be stable.

[286SX]

Someone please post some information about what the Skew settings do and when you might need to change them and in what increments, upper/lower bounds, etc.

I would also like to learn if any of the PLL voltages are actually useful on the i7 platform. I've seen people post that they are using core and/or QPI PLL voltage changes in some cases but they didn't post any info that would tell you that they scientifically determined that they actually needed the PLL changes to be stable.

I played around with Skew and Amp last night. I am stable at 4.0 @ 1.3125v, so I figured I would work on 4.1Ghz.

I set CPU voltages starting from 1.3125v - 1.35v. I set QPIv at 1.225v-1.5v, and set PLLv between Auto-1.96. If the system crapped out during Prime blend, I would go in and set both skews at 100ps. [My P6T manual says that I should adjust both skews at the same time.] If that died, I would set the Amp at 800mV. On my system, there is no BSOD, only reboots when something goes wrong. I have to assume this is CPU related and not power or RAM. My guess is it needs more voltage.

I did read somewhere that as CPU speed increases past 3.8Ghz, the amount of voltage needed increases faster than linearly. But I can't find the article again... Doh.

So far, nothing past 4.0Ghz is stable in prime or linx. I did try one run at 1.375 with no skew or amp changes. It ran fine in prime (with no reboot) - but got hot enough to throttle one core, so I turned it off.

Overnight I set BCLK to 201 [X20] @ 1.3125v and ran prime [small, not blend] for 8 hours.

I am beginning to think that skew and amp are snake oil, and everything is dependent on voltage...

-td

[DavidNJ]

Is this an overclocking forum or a Subaru forum?

[JWMc]

Is this an overclocking forum or a Subaru forum?

LOL, I take no responsibility for the car talk exceeding one post and one reply. But yea I have noticed a fair amount of Subaru avatars lately though.

I may change mine once my upcoming computer chassis is looking good. 286sx took it straight to PMs already and I'm about to reply to his.

To get back on topic: my 190x21 @ 1.312v is of course Prime blend stable, almost 8 hours now and still going. I hope I can stomach the VTT I may need to really push this thing all the way on water and I also hope I can use much less of it with 1GB modules and/or HT off.

[286SX]

I still have no idea what skew and amp do.

After 4100hz, I believe that they don't really help much. And you still have to have high enough voltage to make it work.

Here is my example:
4104 [216X19] - CPUv 1.31875; PLLv 1.84v; QTIv 1.33125 - ran LinX without reboot, but it was HOT, so I reduced voltage to:
4104 [216X19] - CPUv 1.3125; PLLv 1.84v; QTIv 1.33125. This caused LinX reboot. So I changed skew and amp to see if it would work.
4104 [216X19] - CPUv 1.3125; PLLv 1.84v; QTIv 1.33125; CPU Skew 100ps; IOH Skew 100ps; CPU Diff Amp 800mV. Still crapped out in LinX.

Maybe more skew differentials would work. I'll try to squeeze in some tests on this later.


Also, I updated my max OCs HERE.

4085 [215 x 19] @ CPUv 1.3125; PLLv 1.8; QTIv 1.30. This is only LinX 20 tested, not Prime95. [98/97/93/91]
4066 [214 X 19] @ CPUv 1.3125; PLLv 1.8; QTIv 1.30. This is only LinX 20 tested, not Prime95.
4047 [213 X 19] @ CPUv 1.31875; PLLv 1.8; QTIv 1.3. Also LinX 20 tested, not Prime95.
4020 [201 X 20] @ CPUv 1.31875; PLLv A; QTIv 1.225. Prime Small 8 hours.

Anything around or over 4.1Ghz requires better cooling than I have [Cooler Master V8]. I think it would go higher with better cooling - especially since I can hit 4.1 with under 1.325v.

Should've bought the bigger cooler! Oh well. I like the way the V8 looks, and I can run close to 4.1Ghz on it...

-td

[roller11]

I have a gigabyte EX58 board and it has some skew options as well....I was wondering the same thing.

I played around with skew options in BIOS, no improvement.