Intel Core i5 & i7 Sandy Bridge Overclocking and Feedback

[Darkleoco]

Just out of curiosity should I have my memory clocked at 1333 while finding a stable OC then bring my memory back to 1600?

 

[cadaveca]

No, set up the system how you want it to be, and test with it as you would use it.

For Linkpack, you want to use one that supports AVX extensions.

 

[Darkleoco]

No, set up the system how you want it to be, and test with it as you would use it.

For Linkpack, you want to use one that supports AVX extensions.

Should I go back and set my memory at 1600 or let Prime run through ad just set it afterwards?

 

[Chicken Patty]

Set it now, and run it again.

 

[Darkleoco]

Set it now, and run it again.

Running it again with memory at 1600, 1.35v and hopefully it will be stable enough to get in some BF3 later and see if I get a performance increase out of it.

 

[Chicken Patty]

Should yield you a bit of a gain at those clocks.

 

[Darkleoco]

Should yield you a bit of a gain at those clocks.

What do you consider a bit? 5-10 FPS? If it is anything less than that I might just not bother OC'ing at all.

Just crashed at 1.35v after like 20-30 minutes of Prime95 *sigh* now bumping to 1.37 in the hope that it will stabilize finally.

 

[Chicken Patty]

To me 1.37-1.38 is fine. Under 1.4v from what others have posted.

As far as FPS increase I don't know, but I know Crysis 2 runs a lot smoother for me.

 

[Darkleoco]

To me 1.37-1.38 is fine. Under 1.4v from what others have posted.

My only concern is my temperatures though, I was maxing 75 on at least 1 core at 1.35v and 1.37 is certainly going to be significantly worse. :/

 

[cadaveca]

my max voltage comfort zone is built around the differences in temperature between cores. I like to keep that difference within 10c...you should notice that the temperature difference between cores increases as you increase voltage.

That is another wy to tell if your chip is a good one or not...good ones tend to have core temps very lcose to each other, while the same cannot be said for the bunk chips I've had.

 

[Chicken Patty]

There's still some room, stay at no more than 80*. Plus, once done testing you probably won't see temps that high again z

 

[cadaveca]

There's still some room, stay at no more than 80*. Plus, once done testing you probably won't see temps that high again z

Yeah, stress testing I hit 85-90c, no problem, but gaming, very rarely do I get over 70c. Power consumption is also much lower while gaming.

 

[Chicken Patty]

Yep, so he should be fine as far as temps.

My chip, I see about 3-4*c difference between cores at 4.5 GHz/ 1.36v full load. Is that good?

 

[Darkleoco]

Yeah, stress testing I hit 85-90c, no problem, but gaming, very rarely do I get over 70c. Power consumption is also much lower while gaming.

My cores are usually really close except for one.

70
75
75
72

Never had a core off by more than 5 degrees though.

Is that good?

 

[PaulieG]

my max voltage comfort zone is built around the differences in temperature between cores. I like to keep that difference within 10c...you should notice that the temperature difference between cores increases as you increase voltage.

That is another wy to tell if your chip is a good one or not...good ones tend to have core temps very lcose to each other, while the same cannot be said for the bunk chips I've had.

I've heard this before, but curiously, my 2 best chips out of a dozen both had 9-10c differences between the coolest and hottest core, and one of my worst chips only had a 6c difference. Weird.

 

[Chicken Patty]

the complete opposite.

 

[cadaveca]

I've heard this before, but curiously, my 2 best chips out of a dozen both had 9-10c differences between the coolest and hottest core, and one of my worst chips only had a 6c difference. Weird.

I use all of the possible methods...core temp differences, power consumed via the 8-pin, max multi, and voltage used, to deem a chip good or bad. It's not just one thing, but rather a combination of all of the above.



I wish we had more insight into how Intel is binning these chips, but with only one HT CPU on the market, that's kinda hard. Diabling HT makes no difference in power consumed or temps for me, so I am kinda lost even though I've binned 24 chips now.

I think i have a good idea, but damn if there isn't always a wrench in the gears...

 

[Chicken Patty]

So what would you guys say aids in knowing which chip is better binned? Multi?

 

[PaulieG]

I use all of the possible methods...core temp differences, power consumed via the 8-pin, max multi, and voltage used, to deem a chip good or bad. It's not just one thing, but rather a combination of all of the above.



I wish we had more insight into how Intel is binning these chips, but with only one HT CPU on the market, that's kinda hard. Diabling HT makes no difference in power consumed or temps for me, so I am kinda lost even though I've binned 24 chips now.

I think i have a good idea, but damn if there isn't always a wrench in the gears...

So what would you guys say aids in knowing which chip is better binned? Multi?

I've learned about half as much about Intel's binning of SB chips compared to Nahalem. They were far more predictable. Hell, I haven't even been able to say that chips with a higher multi ceiling run cooler at any vcore than a lesser chip. My 54x chip had one core that ran warmer than 5-6 other chips I had. Anyway, I've had about as much fun with Sandy as I care to. Just running one SB in a family rig for now.

 

[cadaveca]

So what would you guys say aids in knowing which chip is better binned? Multi?

Yeah, multi is the real thing. Then, of course, you get two chips with the same max multi...VID doesn't tell you much, unless you also look at how much current the CPU eats @ that voltage.


Also, we can look at how votlage additions allow the chips to scale...lots to bin by, really.

As paulie says, its a complicated thing. Every time you think you foudn the "golden rule", you get another chip that totally breaks all the rules.

That's what makes clocking and binning chips fun, to me, anyway. SB-E is what I'm thinking of now though...1155 was 2011, and 2012 is fast approaching.

 

[Chicken Patty]

Well, I agree with Paul, S1366 i7's were far more predictable.

I miss my i7 920 D0. 4732 MHz on air, in Miami!

 

[Darkleoco]

I think I might finally be stable at 1.37v X_X gotta let Prime run longer though because I honestly can't be bothered doing Linpack lol

 

[Chicken Patty]

let us know how it goes bro.

 

[PaulieG]

Well, I agree with Paul, S1366 i7's were far more predictable.

I miss my i7 920 D0. 4732 MHz on air, in Miami!

Was that my old chip? Can't remember. If it was, it was a hell of a good chip.

 

[Chicken Patty]

Yes it was, thing was a beast!