Core i7/X58 Overclocking Thread

[sonofander]

I don't seem to have issues running my uncore higher that qpi either. Actually my more stable overclocking runs have been with uncore higher than qpi...

[SteveRo]

My 1st pc vantage run - -

[JPQY]

SteveRo..what a setup with all these SSD's ...i hope you want to try my Chess test!!

Just send you a PM.

JP.

[Drwho?]

That's some excellent observations there dejanh.

So what we need is lower QPI multis to really push Core i7 to it's full potential.

I kinda miss the independent memory handling of nVIDIA chipsets.
The limited multis of intel chipset(and i7's mem. controller) are somewhat frustrating.

How you got your Corsair Dominators to 1800mhz baffles me; I cant even get an extra 50mhz out of my 1600MHZ kit.

working on the lower multipliers ... already.

for your 1600Mhz kit, try to increase the QPI/MemCtrl voltage a little. (at your own risk .. lol)

who?

[rge]

EDIT: Dr. Who is the lower QPI multi something that intel needs to work out first or just bios implementation? also mem 9 multi in addition to lower QPI multis is on my GB wish list, so I can run my mem at stock 1600 or little over instead of trying to OC to >1800 with 10 multi...emailed GB with my wish list...but dont know what is possible and what not.

I may run my 24/7 at 4.2, if stays stable, 1 hr so far, still at reasonable vcore. Mine scales at 1 notch higher vcore for every ~23mhz (1 mhz increase in bclk with 23 multi). Also raising cpu pll and ioh core couple notches seems to help on mine. rest settings on pic.

4.2 prime 1h so far with 1.41 bios, 1.376 load.
4.1 15hrs stable 1.38bios/1.34 load
4.0 12hrs stable 1.35 bios/1.31 load

[sonofander]

good job RGE
4.2 with HT on @ 1.4v is very impressive!

[Bootsy]

What speed are you running your memory?
Are you still priming?
Very interested in a stable 4.3 oc

Mem runs at 930mhz(12 hours memtest stable) these Gskill Blacks are great!! Prime crashed afte 1 hour 15 minutes. I am still working on it as I dont think its just a voltage issue. I wll get 4.3 stable.

[sonofander]

nice OC Bootsy!
what version of memtest are you running?


also, does anyone know where to get a list of the order that P95 runs its tests in?

[Hornet331]

working on the lower multipliers ... already.

damn i bought my Ci7 to eraly.

I hope we see those with westmere.

[rge]

good job RGE
4.2 with HT on @ 1.4v is very impressive!

My cpu seems to OC pretty well with low vcore, but it runs hot. I just hosed my 4.2 run, got impatient and ran linpack at same time as prime, it froze at 10th pass of linpack and little over 2 hrs prime. But glad I did, if it cant handle both at same time, it usually reboots at some point, even at 8 hr piss me off point. So up 1 notch vcore and run again.

[M3kk]

How about this, with air.




http://valid.canardpc.com/show_oc.php?id=461672

[rge]

^^That is pretty sweet on air, and nice low vcore.

[Splave]

are you outside?

[SteveRo]

m3kk - wow - excellent - how long can you prime with that?

[rge]

are you outside?

I have not tried the cold yet. But he could easily be inside. In my sig, though i am on water, mine will do 4600, sub 9 spi at 1.49 cpuz, and that is inside with 26C ambients.

In fact I will let u know in hour or so on mine....but I am betting cold does very little, especially after seeing some on ln etc.

EDIT: ok room temp boot max is 4670 at 26C with 1.62vcore, even 1.66vcore did not help (had vtt 1.46, cpu pll 1.98, IOH 1.18), forgot to get screen of it, but pic is 4600 at room temp 26C with spi at 1.5, and validate here:
http://valid.canardpc.com/show_oc.php?id=462368 room temp

EDIT 2:Outside in 5C, open case, box fan up against computer blasting 5C air through my case forcing air through my rads at ?3x air speed of my yates, let set for 30 mins. Max boot was 4760 or 90 mhz gain by going out in cold, ran at 1.64 and 1.66vcore, but would not go 1 baseclock higher (failed to load windows), all settings same as room temp above. (cpuz says 1.68 when 1.66 vcore in bios with loadline...somewhere around 1.6 sensors or volts gets inaccurate.)


so 90 mhz gain...more than I thought, but much less than 200+mhz gain of my E6850 by taking it out in cold.

[sonofander]

My cpu seems to OC pretty well with low vcore, but it runs hot. I just hosed my 4.2 run, got impatient and ran linpack at same time as prime, it froze at 10th pass of linpack and little over 2 hrs prime. But glad I did, if it cant handle both at same time, it usually reboots at some point, even at 8 hr piss me off point. So up 1 notch vcore and run again.

it kills me... I need to set 1.475v bios to be prime stable at 4.3 w/ HT disabled! doesn't look like I got the best clocking chip

at least my temps aren't bad

[dejanh]

Please explain how this is possible then: http://www.xtremesystems.org/forums/...79&postcount=2

2Ghz Ram, 4Ghz Uncore but only 2.5ghz QDI speed with 143Mhz BCLK x18 QDI multiplier.

Has anyone else confirmed that QDI has to be faster than Uncore to POST?

If you ready closely you will see that I said running QPI at or past 4.0GHz results in no POST, not the other way around I added some more information to my original post as well, to make this more clear. Furthermore, I added a note that addresses the example you gave.

Overall, the idea behind my testing was to discover limitations of Core i7 920 and i7 940. Core i7 965 EE certainly has more flexibility in that you should be able to keep your BCLK (and therefore QPI) lower while running better memory due to unclocked multipliers. Furthermore, the idea behind my testing was not to cripple the chip features to get higher clocks (such as turning off HT/cores, etc.) or to be running voltages that are dangerously close to absolute maximums (e.g., 1.5V+).

Some may consider 1.5V safe, but I'd be betting those chips will not be surviving very long of they are run most of the day over extended periods of time. In my opinion, if I cannot stay below 1.46V the clocks I am demanding are too high. But, only time will show.

[M3kk]

are you outside?

No, the ambient temp was ~15 oC.
I will try with open window when will have -10 outside .

m3kk - wow - excellent - how long can you prime with that?

No prime at this speed .

[sonofander]

If you ready closely you will see that I said running QPI at or past 4.0GHz results in no POST, not the other way around I added some more information to my original post as well, to make this more clear. Furthermore, I added a note that addresses the example you gave.

Overall, the idea behind my testing was to discover limitations of Core i7 920 and i7 940. Core i7 965 EE certainly has more flexibility in that you should be able to keep your BCLK (and therefore QPI) lower while running better memory due to unclocked multipliers. Furthermore, the idea behind my testing was not to cripple the chip features to get higher clocks (such as turning off HT/cores, etc.) or to be running voltages that are dangerously close to absolute maximums (e.g., 1.5V+).

Some may consider 1.5V safe, but I'd be betting those chips will not be surviving very long of they are run most of the day over extended periods of time. In my opinion, if I cannot stay below 1.46V the clocks I am demanding are too high. But, only time will show.

what IS your setup? what speeds are you running?

[btdvox]

it kills me... I need to set 1.475v bios to be prime stable at 4.3 w/ HT disabled! doesn't look like I got the best clocking chip

at least my temps aren't bad

Whats your batch #?

[dejanh]

what IS your setup? what speeds are you running?

I already listed my current settings several times...

http://www.xtremesystems.org/forums/...&postcount=880

I can also run it with Turbo on without any issues, though I do have to bump the voltage up slightly.

I'm on R2E 0804 BIOS (beta I believe), Core i7 940 w/ CoolerMaster V8 heatsink, 6GB Corsair Dominator 1600MHz triple-channel...

[sonofander]

Whats your batch #?

3836A514

I already listed my current settings several times...

http://www.xtremesystems.org/forums/...&postcount=880

I can also run it with Turbo on without any issues, though I do have to bump the voltage up slightly.

I'm on R2E 0804 BIOS (beta I believe), Core i7 940 w/ CoolerMaster V8 heatsink, 6GB Corsair Dominator 1600MHz triple-channel...

maybe you should just list it in your sig

[dejanh]

3836A514




maybe you should just list it in your sig

I'm too embarased to post it...still running a single 9600GT, waiting for the 55nm NVIDIA GTX parts I should really put my rig under water too, at least the CPU and chipset. My ambient temps are very high as I like my room warm (21C - 23C normally, around the computer it gets quite a bit hotter too as it is in a corner with no air circulation). I think this is limiting my OC as well as I get up to 91C using 1.46V @ 4.14GHz. I was first overclocking with the stock cooler and I noticed that I could get higher clocks with lower volts using the V8, which leads me to think that going water would be even better (i.e., even higher clocks with lower volts, or at the very least lower volts for same clocks).

[theseeker]

How about this, with air.




http://valid.canardpc.com/show_oc.php?id=461672

Very impressive! I was at 4.6 Prime stable for about two hours until I ran into a loop problem with my H20 setup. Now I seem to be fine at 4.4 but this CPU is a hot tamale!

[unclewebb]

Just a quick fix to RealTemp so the new Load meter is more accurate. Thanks rge for pointing that out.
The About box will show you what version you have.

http://www.fileden.com/files/2008/3/...alTempBeta.zip