SB Overclocking Thread

[reb0rn]

Even if CPU degrade for 98% ppl it will be ~100-200Mhz or need to bump a voltage by 0.02v or something like that, also it might be normal for CPU after 2-3 weeks of use to ask for 0.02v more, call that behavior as you like, but its perfectly normal!

 

[slag]

If I didn't want to overclock I would not have purchased the 2600k. 1.35v is way too conservative. The nay-sayers were out in full force when I was putting 1.55-1.6 through my E6750, as well. That rig put up with that for years and is still running 24/7 in my basement.

You mean Command Bunker?

 

[Diogenes2]

.....On xtremesystems forum there is already users who have had their cpus degrade (2 days I think it took) running close to 1.5V. ....

Could you point to one or more of these discussions? Sometimes there is more to an overclocking failure than CPU voltage..

If people want to be safe, they shouldn't overclock.

I'm personally not going to be running in the 1.5 range due to heating and power considerations, but if Intel says the VID can be as high as 1.52, that means the silicon can handle it.

You can't ( accurately ) compare SandyBridge to other cores because it is a totally new fab.

We can assume they have enhanced the process in some way that yields the numbers they have published .

...
http://en.wikipedia.org/wiki/Sandybridge

Intel itself demonstrated a Sandy Bridge CPU running stably overclocked at 4.9Ghz on air cooling (Details not yet revealed)

You think they might have done this on 1.35 or less ?

 

[Arkainium]

What kind of PLL is everyone running? At 4.6 i'm running 1.4v. What the hell does PLL even do?

Makes you wonder why Intel lists 1.71V as the minimum for this in the datasheet. Are there any adverse effects from undervolting other than possible instability?

 

[sgrinavi]

Any words of wisdom on i7-2600k batch numbers ?

 

[Draax]

You mean Command Bunker?

My old 775 system is now the brains of my CNC milling machine, which I built in my basement. My main rig resides upstairs, in my office, ... Well until my wife changes her mind.

 

[M1A]

with the latest issues with SB I am glad I did not jump in, will wait till next gen comes out.

 

[Diogenes2]

Nice overclock you have there.. Thanks for sharing.

 

[Peter Trend]

Did you see this?

Have you all seen the recall article on Sandy Bridge??

http://news.cnet.com/8301-1001_3-20030048-92.html

Also on Anandtech's front page.

Looks like the P67 chipsets are defective.

http://forums.anandtech.com/showthread.php?p=31180520#post31180520

 

[Dadofamunky]

You're a TAD late on that.

 

[tviceman]

Just thought I'd drop in and share my OC numbers with everyone else.

With the P8P67 Pro, 2600k, and Noctua NH-U12P I am completely stable @ 4.6ghz with cpuid reporting vcore as 1.312v under 100% load, and the highest temp reached out of the 4 cores is 67 degrees celsius. I also have hyperthreading disabled.

EDIT: I take that back. Apparently 10 minutes of Crysis Warhead is more demanding than a few hours of Prime95. I'll try increasing the LLC.

 

[Thor86]

I'll say it again, Prime95 needs "custom" testing using 4096k size FFTs and all ram. If it can pass this in a days worth of running, then you should not run into any other stability issues.

 

[tviceman]

I'll say it again, Prime95 needs "custom" testing using 4096k size FFTs and all ram. If it can pass this in a days worth of running, then you should not run into any other stability issues.

Until it appears prime95 is no longer a valid method for testing system stability with these new chips. I ended up keeping LLC at medium and upping the offset by .005v. At full load my vcore now hits 1.328 but I'm stable at 4.6ghz now.

 

[Beace]

Finally got my machine up and running after a lot of problems with a bad motherboard.

I've only done some basic OC for now, sitting at 4.4 GHz at 1.25 Vcore. The cores seem to max out at 55-60C running IntelBurnTest using a CM 212+ cooler.

My goal was around 4.3-4.4 GHz, but seeing how easy this has been going I might try for 4.6 or 4.7 GHz. A stable 4.4 GHz at 1.25 Vcore seems to imply I got a very good chip, or at least that's the impression this thread has given me.

 

[Castiel]

I've always found BC2 a good benchmark for testing cpu stability

 

[Axon]

I've always found BC2 a good benchmark for testing cpu stability

Zing!

 

[Peter Trend]

Is anybody else getting significant temperature differences between cores at full load and idle? Think my HSF might be badly mounted, I'm at 4.5GHz with 1.38Vcore and getting 65-67C, 72-74C, 72-74C, 73-77C in Core Temp is that normal?

None of them are great temps anyway, I know. I should try dropping Vcore soon as it's 100% stable now, running BOINC and f@h 24/7 for ~3 days. PrimeGrid got it hotter than f@h, too (but I overclocked my GPUs at the same time).

 

[Hogan773]

I've been playing with this today....now running a 44x on Auto with the ASRock Extreme 4 and Vcore is around 1.26. so my questions

1) I've got HWMonitor, CPUID and the ASRock Extreme Tuning utility all open on the screen. All the VCores roughly match. Any reason not to believe them (I'm a little surprised that 44x at 1.26 on Auto is so easy). I opened CoreTemp and its voltage readings are different, but its last version is September way before SB even came out.

2) If my ASRock is being reasonable with voltages on AUTO, any reason to move to Fixed or Offset mode? Right now I can still throttle back to 0.96V and 16X at idle, and my understanding is if I type in "1.25V" Fixed, it will sit there even at idle.

3) What IS Offset mode. Is it basically AUTO but you can specify even more extra voltage? I don't understand what it is and the ASRock manual is about as circular as it can get (Offset Mode: Allows adjustment of Offset Mode parameter - gee thanks)

 

[smakme7757]

This is my highest overclock so far 5559Mhz.

http://valid.canardpc.com/show_oc.php?id=1632365

 

[Peter Trend]

This is my highest overclock so far 5559Mhz.

http://valid.canardpc.com/show_oc.php?id=1632365

Congrats! Some hefty voltage there though, makes me nervous :\

 

[smakme7757]

Congrats! Some hefty voltage there though, makes me nervous :\

It's not as bad as some though. Some chips need 1.60v+ to get 5.5Ghz. Either way once you get over 1.55v it's already too much Vcore .

As strange as it sounds my 24/7 overclock is stock . I even run Gigabytes Dynamic Energy Saver 2 as well. I just don't need it to be overclocked as it's fast enough at stock. Once i need more speed i'll crank it up

 

[Beace]

Is anybody else getting significant temperature differences between cores at full load and idle? Think my HSF might be badly mounted, I'm at 4.5GHz with 1.38Vcore and getting 65-67C, 72-74C, 72-74C, 73-77C in Core Temp is that normal?

None of them are great temps anyway, I know. I should try dropping Vcore soon as it's 100% stable now, running BOINC and f@h 24/7 for ~3 days. PrimeGrid got it hotter than f@h, too (but I overclocked my GPUs at the same time).

I'm observing the same, whether it's normal or not I'm not fully certain. At full load, I'm seeing roughly: 58C-60C-63C-57C, on a 2600K.

I'd definitely recommend trying to get your Vcore a bit lower, most people seem to feel around 1.38-1.4 is absolute max for an everyday OC.

 

[Accord99]

Is anybody else getting significant temperature differences between cores at full load and idle? Think my HSF might be badly mounted, I'm at 4.5GHz with 1.38Vcore and getting 65-67C, 72-74C, 72-74C, 73-77C in Core Temp is that normal?

Those temperature differences seems universal in Sandy Bridge. Assuming the first core is the one closest to the IGP, the temperature differences seems reasonable because the middle cores are surrounded on both sides by heat producing cores, while core 1 is beside the usually gated down IGP.

 

[Salatheon]

My 2500k running at 4.7Ghz has strange temps too, at idle all 4 cores sit at ~40-45c (In Australia summer with no Aircon atm), then under load (Prime95) i'll get Cores 0-1 at ~70-72, then core 2 can be as much as 10 degrees C hotter and core 3 at around 77c.

Also, does anyone else notice once you go from full load to idle the temps drop INSTANTLY by 20-25 degrees Celsius? Because mine do..

Heatsink-> CoolIT Vantage ALC CPU Cooler (self contained liquid cooling like the Corsair H70)
CPU-> i5 2500k @ 4.7Ghz with 1.35v

 

[Hogan773]

My 2500k running at 4.7Ghz has strange temps too, at idle all 4 cores sit at ~40-45c (In Australia summer with no Aircon atm), then under load (Prime95) i'll get Cores 0-1 at ~70-72, then core 2 can be as much as 10 degrees C hotter and core 3 at around 77c.

Also, does anyone else notice once you go from full load to idle the temps drop INSTANTLY by 20-25 degrees Celsius? Because mine do..

Heatsink-> CoolIT Vantage ALC CPU Cooler (self contained liquid cooling like the Corsair H70)
CPU-> i5 2500k @ 4.7Ghz with 1.35v

Yep mine go from 58C at load down into the high 20s in like 2 seconds. I guess that big metal 212+ is sucking all the heat out quickly!