Intel Core i5 & i7 Sandy Bridge Overclocking and Feedback

[johnnyfiive]

Trying to find the highest it will go before needing 1.3v

 

[johnnyfiive]

http://i59.photobucket.com/albums/g307/lubetek/i52500k5GHz1424vidle148vload.jpg


This is not stable though. I can run IBT 24 hours straight with no errors but I can't surf the web without getting a BSOD. I'm not really sure what I'll do to make it 100% stable. As of right now I'm back at 4.7GHz which is stable for sure. I'm very new to this platform though so given time I'm sure I will work it out.

Whats you're PLL and PCH voltage?

 

[johnnyfiive]

Guess I'll stress this more tonight. Would be awesome to get this stable using 1.3v for 4.7

 

[Undead46]

Stable at 4.5GHz, 1.35V (1.36V during Load), 1.7PLL.
100 Passes of Intel Burn.
4 Hours of Prime95.
63C High.

Finally got my computer to stop hard crashing..

 

[PaulieG]

Can't believe this is actually stable...temps are still high after the test because WCG is running in the background.

 

[heky]

Nice clocks guys. Its getting harder and harder to wait for my 2600K, just staring at my P67A-GD65.
@WhiteNoise
You are crashing at idle becouse you are not getting enough voltage(or current) at idle. Try up-ing either one.

 

[WhiteNoise]

Whats you're PLL and PCH voltage?

they are set to auto.

And its official. I'm rock solid stable at 5GHz. Been testing, benching, stressing, gaming, watching videos...and no BSODs.

I'll post screenies later.

Thanks goes out to cadaveca for telling me how system agent voltage helped him. I brought mine up and BAM! all is well.

Now when I feel like it I'll have to start lowering VCore and see what happens.

I also selected SPD for memory which detected it as 1066 and less voltage. I upped the voltage to spec for the memory and left it at 1066. So I now have to put it back up to 1600 and se what happens.

I'll do that first before touching vcore.

Anyways once I decide on my final settings I will post them up. But for now I'm stoked to be stable.

 

[heky]

Nice. I am glad you figured it out. Please post some pictures when you can, and the setting you have set. I have the same board, just waiting for my 2600K. Also can you tell me what bios version you are using?

 

[WhiteNoise]

UPDATE: Okay so I went back into the BIOS and placed my memory at 1600 and set everything to the memory's specs. Once within windows I crashed within a couple minutes.

So....my memory is the problem. I'm glad cadaveca brought the memory up because I wasn't thinking anything was wrong with it since I wasn't overclocking the memory but as it turns out this memory and this board do not jive.

I'm going to do some research tonight and buy new ram tomorrow. If I have to I'll buy top end kit.

Nice. I am glad you figured it out. Please post some pictures when you can, and the setting you have set. I have the same board, just waiting for my 2600K. Also can you tell me what bios version you are using?

I'm using 1.9 (the latest one). Once I have my new memory and I do some testing I will post my bios settings.

 

[heky]

There is a 1.10b6 bios on the Msi forums, and it is supposed to fix many memory issues. Maybe you should try it, it might fix your issues.

 

[WhiteNoise]

There is a 1.10b6 bios on the Msi forums, and it is supposed to fix many memory issues. Maybe you should try it, it might fix your issues.

Funny. I just bought my board this past Monday and updated the BIOS and since then another one released. Thanks man.

I will try this out tonight and see what happens.

There is a 1.10b6 bios on the Msi forums, and it is supposed to fix many memory issues. Maybe you should try it, it might fix your issues.

So after looking at the MSI site I see that the BIOS you are talking about is not official yet. 1.9 is the latest official BIOS. 1.10b6 is a beta BIOS.

That said the memory controller is built into the cpu now and the cpu only supports 1066 and 1333 from what I have read. So tomorrow morning when I wake up I will head over to Fry's and pickup some good 1333 memory. Now I'm not saying faster memory can't be used but in my case using 1333 will be my best bet. Or so I hope.

 

[cadaveca]

SandyBridge deals best with 1.5v or lower mem. 1600mhz 1.5v+ mem, seems to need IMC volts boost, but 1.35v 1600 is fine with IMC @ stock.

Of course, I'm using 4 sticks, will play more with 2 sticks and see what else is up.




Eventually wil lbreak down the wall, gonna pick up some new cpus this weekend most likely, nothing left much to explore board-wise.

 

[WhiteNoise]

Yeah I'm running my 1600 1.65v at 1333 and 1.5v and all is well but it bothers me that I can't run full speed on the memory.

I'm tempted to look for 1600 that runs at 1.5v but I'm betting I'll be happy with 1333. I found some Corsair xms3 2x4gb that should do the trick.

 

[Undead46]

Stable at 4.5GHz, 1.35V (1.36V during Load), 1.7PLL.
100 Passes of Intel Burn.
4 Hours of Prime95.
63C High.

Finally got my computer to stop hard crashing..

This was fail...
Meant to say 1.335VCore*
Hitting 1.336V under full load.

So, volts are pretty low.

 

[PolRoger]

Now testing "cadaveca's" high current overclock method...

Using Vcore offset, set to 1.2v in BIOS, 0% LLC, 160% current, 47x, 4 dimms populated @ 2133 speed.

PLL voltage overide disabled, Speedstep, EIST and all C-states enabled. Dynamic VID has vcore ranging from ~1.28v to 1.30v load (varies with stress/load/droop).

 

[cadaveca]

can you drop initial votlage to 1.175v or lower?

how about playing with VRM frequency?

Those options are there for a reason!

 

[PaulieG]

SandyBridge deals best with 1.5v or lower mem. 1600mhz 1.5v+ mem, seems to need IMC volts boost, but 1.35v 1600 is fine with IMC @ stock.

Of course, I'm using 4 sticks, will play more with 2 sticks and see what else is up.




Eventually wil lbreak down the wall, gonna pick up some new cpus this weekend most likely, nothing left much to explore board-wise.

I've not seen this with 2x4GB sticks. Probably only occuring when all dimm slots are filled.

 

[Frizz]

Hmm I'm going to return my 1.65v ram tomorrow, which 1.5 kit should I got with Corsair or Gskill?

http://www.itestate.com.au/pages/pr...aces?BeanName=productDetailAction&pcode=08742

http://www.itestate.com.au/pages/pr...aces?BeanName=productDetailAction&pcode=08748

GSKILL model seems to have lower timings, but then again they are the same price so I am not quite sure how either one will affect my system.

 

[WhiteNoise]

Been playing with voltages since yesterday. I reached stability earlier today but only by using higher voltages. Once I hit stability I have now been slowly working one voltage at a time down. I hope to find my final settings this evening. Will post all my settings once I do.


5Ghz is a reality for me so far.

Hmm I'm going to return my 1.65v ram tomorrow, which 1.5 kit should I got with Corsair or Gskill?

http://www.itestate.com.au/pages/pr...aces?BeanName=productDetailAction&pcode=08742

http://www.itestate.com.au/pages/pr...aces?BeanName=productDetailAction&pcode=08748

GSKILL model seems to have lower timings, but then again they are the same price so I am not quite sure how either one will affect my system.

I returned my Patriot Viper II Sector 5 G-Series 1600 last night and bought Corsair Vengeance 2x4GB 9-9-9-24 and so far its been awesome. It's also intel certified to work with i3, i5 and i7 as well as X.M.P.

Corsair Vengeance DDR3 1600 8GB (2x 4GB)

 

[cadaveca]

I've not seen this with 2x4GB sticks. Probably only occuring when all dimm slots are filled.

I have one 5.3ghz and one 5.6ghz chip. Seems CPU dependant. Makes me wish I did more exact binning rather than just for multi.

 

[WhiteNoise]

Okay here are my final settings. I'm 100% stable. Vcore is a little higher than I'd like but is 100% required to run this chip at 5GHz. I still plan to try for 5.1GHz and see if I can do it at the same voltages. I won't up my vcore any more. I'm done.

Oh and I plan to run this chip at these settings 24/7. During IBT and using 'high' setting my load temp reached 72c on one core but during regular use the my load doesn't go past 49c.

 

[LifeOnMars]

Wow that is high voltage!! At least temps are good though with it being under water .

I would love to know the length of time these chips will survive. Anyone heard anything else regarding degradation or deaths in our little sandy bridge family?

Cadaveca - A virtual beer on it's way to you as gesture of friendship. Was having a rough day the other day and for some reason your post/points of view raged me Life's too short and all that.

 

[cadaveca]

Cadaveca - A virtual beer on it's way to you as gesture of friendship. Was having a rough day the other day and for some reason your post/points of view raged me Life's too short and all that.

Yeah, sry, that's just me. No worries, water under the bridge, and really, it's no big deal. You guys do know we have a teamspeak server, right?


I killed my first chip back shortly after launch. Not sure exactly what the story was there. At first the chip tested OK by the retailer, so I was sure it was the board. But I tossed another chip in, and it works fine.


Which brings me to my story about binning.

These cpu vary in how they actually perform quite a bit.

We got some chips with really cool cores. With such a large piece of silicon, realtively speaking, it's quite clear that part of the reason these chips clock so well is due to the inactive bits acting as cooling when not in use.

There are a few consistent bits that need to be adressed voltage-wise, like System agent and such, but anything other than that is an outright crap shoot.

In all the cpus I tested, lower stock volts = higher multis, and lower power ocnsumption, but that doesn't mean a chip with a higher multi will actually reach that multi with reasonable volts...

MY lower-multi chip, the one you guys have seen, requires low volts for 4.7ghz. The high-multi chip requires signifigantly more. Current additions have hit the limit this board can provide, and voltage seem the only way.

 

[PaulieG]

In all the cpus I tested, lower stock volts = higher multis, and lower power ocnsumption, but that doesn't mean a chip with a higher multi will actually reach that multi with reasonable volts...

MY lower-multi chip, the one you guys have seen, requires low volts for 4.7ghz. The high-multi chip requires signifigantly more. Current additions have hit the limit this board can provide, and voltage seem the only way.

I've come to much the same conclusion. These are curious chips, but at the end of the day, they are not all that different then previous generations, except I've notice in some chip there is not a linear relationship in multi's and voltage, even with lower multipliers...more than prior generation chips.

 

[cadaveca]

I have played with four different batches now, and within a batch, there seems to be some consistency in quality, other than the multi bits, in such a way you can almost tell what's going on by just popping the chip in, checking volts, and power consumption.


There are chips that do not like bclk adjustments...the low-multi chip won't budge one bit, and prefers to be slightly lower than 100bclk, creating a false max multi, but the high multi chip I've had up to 109...and the high multi chips that I binned earlier didn't like bclk much either...

Each chip has almost endearing qualities...I sold a 59 multi for a good profit, but don't regret it one bit! There are so many ways that these chips can differ, it's really quite interesting to me.