**Official** Core i7 860/870 overclocking thread

[RADCOM]

I've been meaning to start one and the i5 one has given me inspiration. I don't mind correlating the results in a table/spreadsheet. I'm open to suggestions on what constitutes stable. Chronic, namegt, emuexport and a few ohers have obtained good resultt @ around 4ghz so far.
Basic overclocking guide here
XS thread of interest 1 2

[hokiealumnus]

3.8GHz stable; 860 on an EVGA P55 FTW






4.0GHz stable, same setup.







Will have to look at the batch # when I'm at home, I assume you'd like that for your chart?

Thanks for starting this up!

EDIT - Batch # L926B509

[RADCOM]

I've got to 3.7Ghz prime95 stable with 8GB of memory but temps were hovering around 90 degrees. EvGA P55 FTW, ci7 860 8GB gskill ripjaw, Xigmatech S1283

[KURTZ]

@ RADCOM: need to know your voltages and temps ...

[OC Nub]

So far I have 4100mhz LinX stable, have not run an 8hr blend though. I lose video in LinX at 4200mhz. First time it happened was at 4.0ghz so I upped the PCH voltage and seemed to fix the problem. I am unsure what some of these voltages are safe to run at for 24/7 use.

[miahallen]

Just playing around on water cooling tonight, getting a feel for this combo before benching with the cold stuff tomorrow night. First run was nice and easy at 1.45V...22-25C ambient.



Later it got a bit colder so I opened the window (10C ambient) and I was able to give it a bit more juice 1.475V

[unclewebb]

miahallen: Can you try upgrading to RealTemp 3.36? It should do a better job supporting these new CPUs and will report your MHz correctly.

I'm looking for a tester to post or PM me a comparison of RealTemp 3.36 vs CPU-Z when C3/C6 is enabled in the bios while you are in the middle of running a single threaded benchmark like SuperPI. During this test the multiplier should be constantly cycling between the default turbo multi and the maximum turbo multi. I'm hoping to see RealTemp report the average multiplier on the hardest working core while this benchmark is running. TIA for any feedback.

[KURTZ]

miahallen: Can you try upgrading to RealTemp 3.36? It should do a better job supporting these new CPUs and will report your MHz correctly.

I'm looking for a tester to post or PM me a comparison of RealTemp 3.36 vs CPU-Z when C3/C6 is enabled in the bios while you are in the middle of running a single threaded benchmark like SuperPI. During this test the multiplier should be constantly cycling between the default turbo multi and the maximum turbo multi. I'm hoping to see RealTemp report the average multiplier on the hardest working core while this benchmark is running. TIA for any feedback.

Uncle, you need that these tests should be done with the processor overclocked or not?
i've asked to my friends that own this cpu ... so i think the first tests are on the way ..

[unclewebb]

Thanks for the help KURTZ. I was hoping to see the above test with the BCLK at the default of 133 MHz so you don't need to do any overclocking. Enabling turbo allows these CPUs to clock up the multi all by themselves and it should also add additional core voltage automatically if the processor needs it.

I'm just interested to see what RealTemp 3.36 reports when C3/C6 is enabled for the multiplier when not overclocking. The multi on an i7-860 can cycle up to 26X when turbo and C3/C6 is enabled in the bios. That's how Intel designed these things so I just want to make sure that software can correctly report this. I'm not sure if CPU-Z will report 25X or 26X or a bit of both during this test. RealTemp works out the average and should hopefully report a multiplier somewhere between those two numbers when running a single threaded bench like SuperPI. It would also be good to see if RealTemp can correctly recognize these new Xeon chips or if I need to give it a tweak.

[RADCOM]

@ RADCOM: need to know your voltages and temps ...

temp was 90 degrees vcore 1.5 vtt 1.35 ram 1.65 I was just experimenting I'm going to town on it this week.

[xXxDieselxXx]

My final stable clock

[KURTZ]

what is your chip's batch Diesel?

[DAK1640]

I forgot to take a picture of my Prime95 stability but my CPU is crunching WCG 7x24

I compared SP1M to i7 920 at same QPI and the 860 beat it, but just barely.

[xXxDieselxXx]

what is your chip's batch Diesel?

L927B308

Bought from NewEgg on 9/30/2009

[RJR]

Thanks for the help KURTZ. I was hoping to see the above test with the BCLK at the default of 133 MHz so you don't need to do any overclocking. Enabling turbo allows these CPUs to clock up the multi all by themselves and it should also add additional core voltage automatically if the processor needs it.

I'm just interested to see what RealTemp 3.36 reports when C3/C6 is enabled for the multiplier when not overclocking. The multi on an i7-860 can cycle up to 26X when turbo and C3/C6 is enabled in the bios. That's how Intel designed these things so I just want to make sure that software can correctly report this. I'm not sure if CPU-Z will report 25X or 26X or a bit of both during this test. RealTemp works out the average and should hopefully report a multiplier somewhere between those two numbers when running a single threaded bench like SuperPI. It would also be good to see if RealTemp can correctly recognize these new Xeon chips or if I need to give it a tweak.

[IMG][/IMG]

[RJR]

Only played with her for a few hours, folding rig and all, but one of these days I'll have to stop folding on her for a day or two and see what she can do. This is on Air.

[DAK1640]

RJR, you should up your QPI link and re-run SP1M...Go as high as your Ram will allow...maybe 225 x 20 for example...???Your run would be amazing.

[RJR]

RJR, you should up your QPI link and re-run SP1M...Go as high as your Ram will allow...maybe 225 x 20 for example...???Your run would be amazing.

Yes, I really have to sit down and spend a day or two messing with it. It's just this darn Folding addiction, I stop it for a couple hours and start getting the shakes.

[DAK1640]

Yes agreed, this addiction is hard to overcome.

[aamsel]

Only played with her for a few hours, folding rig and all, but one of these days I'll have to stop folding on her for a day or two and see what she can do. This is on Air.

I assume this is on your Gigabyte UD4P?

What batch is your 860?

[RJR]

I assume this is on your Gigabyte UD4P?

What batch is your 860?

Yep,UD4P Batch: L921B884T

[Hockster]

Pertinent stats are in my sig.

CPU is at 3.86GHz, 22X167. I didn't feel like messing too much with ram dividers.
CPU voltage: 1.25
CPU PLL: 1.82
IMC: 1.33
DIMM: 1.60

Temps max at 63-66°C under Prime 95 using the blend test. Hyperthreading is enabled.
HSF is a Zalman CNPS10X Quiet, stock 120mm fan running at 1200-1400 RPM.


Batch# L930B518

[KURTZ]

guys, just a question: what is the SAFE voltages for:
vCC, vTT, vPLL and IMC? thanks

[xXxDieselxXx]

Finally stable at 4GHZ

[xXxDieselxXx]

guys, just a question: what is the SAFE voltages for:
vCC, vTT, vPLL and IMC? thanks

Kurtz, in my case I used this as guidance (This is for eVGA P55 FTW mobo but I'm sure can be applied in any P55 board)

Source

PCIE Frequency Setting to 104MHz. (In some cases this can help maximize BClock)
* QPI Frequency Selection to 4.270 GT/s
* EVGA VDroop Control - Without VDroop
* Bootup CPU VCore - 1.30000V (good for starting point, will likely need to increase for 24/7 stability)
* Eventual CPU VCore - 1.30000V (good for starting point, will likely need to increase for 24/7 stability.)
* DIMM Voltage - 1.65V
* Bootup CPU VTT - 1.400V
* Eventual CPU VTT - 1.400V