Intel Core i5 & i7 Sandy Bridge Overclocking and Feedback

[Nordic]

@james888

What bios are you using? Upgrade to the latest from the MSI site. I think its 1.E

Then,

Cpu base frequency: 100
Multiplier: 45 you can leave it at stock also, and only overclock via turbo multipliers
Internal Pll Overvoltage: Auto
Speed spectrum: disabled
Vdroop control: low-vdroop
Core voltage: 1.35 V
Cpu I/o voltage: 1.09 V or higher, depends on the ram speed and timings
Dram voltage: 1.5V - 1.65V depends on ram again
System agent voltage: 0.93 V
Cpu pll: 1.8 V

You can also leave all the power savings features enabled on the sanybridge platform, so you get low power consumption @idle and high power/speed @load.

I have not upgraded the bios since I got it. I should check and see if there is an update.
As I am a new to this, I would not know, but what I have read says not to use turbo. So your saying I should put the turbo multipliers to my desired overclock. I guess that makes sense. I will try that too.

System agent is pretty high too? And I/O seems low? Weird...but hey, maybe, maybe.

could be due to droop?


I'm going go out on a limb and say you've got an MSI board?

It is an Msi p67 gd53

Unless you have a poor chip, you shouldn't need that kind of vcore for 4.5ghz. Same with your CPU pll. You shouldn't need more than 1.8v here. Were you failing at lower voltages? When you say your temps never exceed 65c, what programs are you using to test?

I use occt and prime 95 to test. I also leave it at that clock speed for a few days just to see if it remains stable.

Also I just tried at 4.6 and it works tested so far. I also have to say thank you for the help. This is exactly what I needed. I will check to see for responses. I will let you guys know how I do with this new information.

 

[Chicken Patty]

Still impressed on how well these damn SB's clock and with low voltage.

 

[heky]

Why can't I have a nice rounded number without having to change blck

Disable spread spectrum in bios and you will have a nice round number.

@james888
Who told you not to use turbo? As i said before, that is the whole point of the sandy bridge architecture, low vots/clock/power usage @idle and high speed @load.

 

[Nordic]

I tried for 4.6 with these new settings. It wouldn't go even at 1.4V. So now at 4.5 again testing at 1.36V. It failed at 1.35.

They didn't really say not to use turbo but said it was not the best way for performance. Even some went as far as to say it was a joke. Last night I tried with turbo and windows would get stuck in boot if I had one of the multipliers set higher than 4.3. I know I can get 4.5 stable on my previous settings so... I don't know what that is.

 

[cadaveca]

They didn't really say not to use turbo but said it was not the best way for performance. Even some went as far as to say it was a joke.

I tihnk there are just some CPUs that prefer non-Turbo, but the majority do prefer it run with turbo on. You'll notice that here on TPU, it was me arguing for Turbo, and everyone else said no Turbo...and then when a couple of users tried my method..they got better clocks, or tmeps, or lower volts?

Today, more people seem to agree with using Turbo now, than earlier this year. It's been an interesting progression. Sounds to me liek you have a dog of a CPU, in comparison to some others...I'm running my review rig CPU @ 4.5GHz with 1.275 or less, usually. those that were making a case of non-Turbo seemingly always needed really high volts..

I guess what I'm trying to say is that you should just do what feels best for you. Really, myself, not being a guy who does folding or none of that, I leave my CPU at stock for gmaing, more often than not. I don't really get much of a NOTICIBLE performance boost with overclocking unless using two VGAs, when it comes to games.


If it works, and is stable..who cares? It's when you start running into issues that you need to try a different approach.

 

[heky]

james888, have you updated the bios? If not, do it, then try the settings above. They should work. I have not seen a 2500K not doing 4.6 and i have built a lot of machines.

 

[Nordic]

I did update the bios. I am starting to think I just have "a dog of a chip." It will start up at 4.6 at 1.35V. Just cant do prime95 for even 5 minutes. At 1.4v it lasted 5 hours. I am now trying to get the lowest volt stable for 4.5. It failed after 3 hours prime95 on 1.36 V at 4.5.

I might end up just settling for 4.4 at 1.35V because it is stable there with the lower volt.

 

[Frizz]

Disable spread spectrum in bios and you will have a nice round number.

didn't work

 

[heky]

didn't work

Hmmm, that is strange. On my board when i disable spread spectrum, i get a round 100, and when it is enabled i get 99.8. Must be something else on the Gigabyte then.

 

[cadaveca]

Must be something else on the Gigabyte then

Many boards are this way. In fact most I've played with sit @ 99.8 MHz, spread spectrum or not. That's ASUS, Gigabyte, Zotac, Biostar, Gigabyte, ECS....

And then the bus doesn't move to 100.0 on quite a few too...you have to choose 100.3, to get 100.1 or 100.2.

 

[Black Haru]

I did update the bios. I am starting to think I just have "a dog of a chip." It will start up at 4.6 at 1.35V. Just cant do prime95 for even 5 minutes. At 1.4v it lasted 5 hours. I am now trying to get the lowest volt stable for 4.5. It failed after 3 hours prime95 on 1.36 V at 4.5.

I might end up just settling for 4.4 at 1.35V because it is stable there with the lower volt.

I ran into a similar wall. I am %100 stable at 4.6 and 1.36v, but to get 4.8 I had to run 1.47v. the one time I tried to run 4.9 on auto voltage it booted at 1.56v.

 

[Frizz]

Many boards are this way. In fact most I've played with sit @ 99.8 MHz, spread spectrum or not. That's ASUS, Gigabyte, Zotac, Biostar, Gigabyte, ECS....

And then the bus doesn't move to 100.0 on quite a few too...you have to choose 100.3, to get 100.1 or 100.2.

Yeah that is true, it is a tad confusing as well although I know it doesn't really affect anything in terms of stability it just gets on my nerves sometimes. Right now I have it set to 100.2 in the BIOS and it rounds the number up with an extra 0001.00mhz on CPU-Z but CPU-Z also reports that the bus speed is 100.0 despite what its set to in the bios . Also there is an issue with the detection under windows 7 and it always reports the CPU at stock speeds, its not a big issue just annoying.

 

[Nordic]

Just to see what the volt would be set at if I set it to auto. At 4.5 it brang me to of 1.352V. Prime95 failed 4.5 at 1.36V. I thought that was interesting. If i raise the multiplier by 1 it did the same thing as it did when I tried to overclock it to 4.4 with turbo. It would get stuck in windows boot. It just begins to boot. It shows a windows logo. It does a little animation, then frozen.

Does anyone know why it would freeze in boot because an overclock? That seems odd to me.

 

[sneekypeet]

not odd at all. its very typical for the way SB CPUs crash when you are either at the multi wall or you need more volts to get it into windows correctly.

 

[heky]

Many boards are this way. In fact most I've played with sit @ 99.8 MHz, spread spectrum or not. That's ASUS, Gigabyte, Zotac, Biostar, Gigabyte, ECS....

And then the bus doesn't move to 100.0 on quite a few too...you have to choose 100.3, to get 100.1 or 100.2.

This just might be, but on MSI boards you can correct it by disableing spread spectrum, and you dont have to put in 100.2 to get the even number.

 

[Nordic]

numbers

This just might be, but on MSI boards you can correct it by disableing spread spectrum, and you dont have to put in 100.2 to get the even number.

I never mentioned it because i know its just something my board does, but disabling/enabling spread spectrum does not change the values to even numbers for me.

 

[manofthem]

I finally got my first Intel CPU, a 2600k with a P8P67 Pro. Put it together yesterday, and today I finally got to mess around with settings. So far, what I have come to is 4.6GHz on 1.328V. I did some initial stress testing but not too extensive yet.

I"ll post back more info as I progress and get more familiar with the BIOS/settings.

I did get to read some of this thread from the beginning and some other spots, and I will continue to do so to see what everyone else has done/is doing to direct me in my quest.

Update:

So I played around a little bit, and this is what I've come to...



4.6GHz at 1.336v according to CPUz on load. How do those temps and numbers look for that speed?

I need to keep playing with it and learning more and more. I want to try to tighten the timings more on the RAM.
I'd like to go higher, but I don't want to go much higher than those temps. Isn't 72.5C the max for this CPU?

Anyway, any info or advice would be welcomed. Thanks a lot.

 

[heky]

I never mentioned it because i know its just something my board does, but disabling/enabling spread spectrum does not change the values to even numbers for me.

You have to disable it, its a well-known fact on MSI boards. I built at least a dozen systems with MSI P67A-GD55/65, and they all had round numbers after disableing spread spectrum.

 

[PaulieG]

I finally got my first Intel CPU, a 2600k with a P8P67 Pro. Put it together yesterday, and today I finally got to mess around with settings. So far, what I have come to is 4.6GHz on 1.328V. I did some initial stress testing but not too extensive yet.

I"ll post back more info as I progress and get more familiar with the BIOS/settings.

I did get to read some of this thread from the beginning and some other spots, and I will continue to do so to see what everyone else has done/is doing to direct me in my quest.

Update:

So I played around a little bit, and this is what I've come to...

[url]http://img822.imageshack.us/img822/9827/linxrun1.png[/URL]

4.6GHz at 1.336v according to CPUz on load. How do those temps and numbers look for that speed?

I need to keep playing with it and learning more and more. I want to try to tighten the timings more on the RAM.
I'd like to go higher, but I don't want to go much higher than those temps. Isn't 72.5C the max for this CPU?

Anyway, any info or advice would be welcomed. Thanks a lot.

Your voltage for 4.6 seems about average for a "decent" chip. However, your temps seem just a little high, considering that those temps are similar to my 5.0 1.376v temps with a Venomous-X. What are your ambient temps like? Maybe your chip just runs a little high, but I'd check that your TIM is spread evenly and the block is mounted correctly.Also, have you tried a slightly lower vcore yet?

 

[Nordic]

You have to disable it, its a well-known fact on MSI boards. I built at least a dozen systems with MSI P67A-GD55/65, and they all had round numbers after disableing spread spectrum.

You and my origianal sources say to disable it. I have not tried overclocking with it disabled. It does not changed the numbers. With or without it on. I am not really sure what to say. It is disabled.

 

[johnnyfiive]

Looks like i'll be joining the sb crew, AGAIN. I was going to go AMD and wait for Bulldozer... but then I realized it wasn't worth waiting for.. unfortunately.

 

[Nordic]

bulldozer

I too was going to go for bulldozer. But well, didn't. A sale I could pass up that saved me $100 was hard to pass.

 

[heky]

Looks like i'll be joining the sb crew, AGAIN. I was going to go AMD and wait for Bulldozer... but then I realized it wasn't worth waiting for.. unfortunately.

And i think you wont regret it! Sandy

 

[manofthem]

Your voltage for 4.6 seems about average for a "decent" chip. However, your temps seem just a little high, considering that those temps are similar to my 5.0 1.376v temps with a Venomous-X. What are your ambient temps like? Maybe your chip just runs a little high, but I'd check that your TIM is spread evenly and the block is mounted correctly.Also, have you tried a slightly lower vcore yet?

Thanks for the advice. I've been messing with it a bit and got things a little better.



I did recheck the cooler and TIM. It didn't look problematic, but obviously I reapplied it, as well as upping the fan speed a bit on the H60 cooler. I got the voltage lowered from 1.336 to 1.312 at load. It shoots up a little to 1.320-1.328 after the testing, but it's still better, which I'm assuming accounts for the lower temps seen. I've only tested it with IBT with 5 to 20 runs. After many fails and bsods, it's semi-stable for now. It's not done yet, but enough for tonight, as I've been messing with it all afternoon and night. Thanks for the tips!

 

[LAN_deRf_HA]

Anybody see anything I could try tightening up here? 4 sticks @ 1866 so I'm pretty confident primary timings won't budge, but what about the rest?