Tips/settings for SB I7 2600K to OC to or above 4.7ghz? 16gb ram?

markm75

Gawd
Joined
Sep 12, 2008
Messages
856
I'm about to get the i7 2600K and the P8P67 Deluxe board.

Yeah i heard about the recall but i should be fine as i dont use many onboard sata ports (just 2, one for OS, one for optical), ill stick with the sata 6Gb ones (which arent affected).

I'd like to shoot for at least 4.7ghz, if not 4.8 or even 5.. i've heard there is stability issues much above 4.7? For 24/7 usage, should i stick with 4.8 or lower?
Should i update to the 1204 bios if its not already on it? I've seen an average of about 1.4 volts for 100x47...

Is getting things OC'd as always.. disable speedstep, c1E, change dram volts to stock, change cpu vcore etc? (thought i read that actually to leave speedstep on this time around?) I saw the one thread on here with a list of things to change off the bat.. do people typically do all of that? (http://www.hardforum.com/showpost.php?p=1036716419&postcount=1)

Finally.. i had 3, gskill 1600GB ram chips in my old board (OCd fine) at 4GBx3.. so i'm adding a 4th since this board is dual channel.. Anyone running 16GB of memory? (i do alot with virtualization so the extra memory is useful)?

I'm getting the H70 cooler.. does everyone just use the thermal grease thats on the cooler, or remove it and add arctic?

Thanks in advance.
 
Last edited:
4.7ghz is about all you will get with the h70. Well thats all i could get she gets realy hot after 4.5ghz.


i was able to take mine to 5.1ghz on the h70 but after an hour or so benchmarking she was in the high 80's c
 
4.7ghz is about all you will get with the h70. Well thats all i could get she gets realy hot after 4.5ghz.


i was able to take mine to 5.1ghz on the h70 but after an hour or so benchmarking she was in the high 80's c

I'd be happy with 4.7, if i can get there, i guess it varies :)

Did you make the changes on the guide posted here.. or only change certain values?

Did you leave speedstep on? It would be nice if it were possible to have the idle speeds get throttled down to save on electricity :)
 
I have an H70 & 2600k @ 4.8Ghz, 1.39v and temps max at 62C at full load after an hour. I did replace the stock fans with gentle typhoons, not sure if that matters. (Also, it's in a silverstone FT02 case with the fans set to exhaust).

Temperature most likely will depend on how much voltage you have to set the chip to.

On sandybridge the processors all vary wildly in terms of what can be a stable overclock, and what voltages are required to get there. It's possible your 2600k couldn't go higher than 4.6, and likewise it's possible it could be stable at 4.9Ghz at 1.38v. It will vary, and you can't tell in advance.
 
Dropping the PLL voltage on mine from 1.77 to 1.7 dropped my load temps by 5C with no loss of stability.
 
Dropping the PLL voltage on mine from 1.77 to 1.7 dropped my load temps by 5C with no loss of stability.

Nice find Forceman! :)

Did the same for me too... mine was at 1.8v at default.

I'm getting 62c running blended prime95 now @4.6ghz
 



I need to check my PLL Voltage in a bit too, might help me out a few degrees.
 
Dropped mine a tad, but nothing crazy. I am enjoy 4.6GHz with 1.29 volts in the bios. I might try 4.7 GHz tomorrow but I just don't feel like going over 1.3 volts.
 
This is unreal. So I take the chip out. Clean it off. Clean the corsair plate etc. Do the reapply. And now I can't even hit 4.8 wo a bsod stop 101 error (voltage I think) even 1-2 hours into prime95. This was all in an attempt to fix the thermal paste which I thought was done wrong, giving me higher temps.

Grrrr I don't get it. Same settings I was 24 hour stable a little over a day a go
 
Dropped mine a tad, but nothing crazy. I am enjoy 4.6GHz with 1.29 volts in the bios. I might try 4.7 GHz tomorrow but I just don't feel like going over 1.3 volts.

I have mine at 4.6 too. Curious do you have the voltage set to manual at 1.29v? I want to understand how to use the offset better. Any thoughts...suggestions etc? Thanks


Elo
 
Tried redoing the paste yet again. Turned it on, Then suddenly the mem ok red led came on the motherboard and or the CPU red led. Couldn't post. took out all but one ram chip n I could. But then the CPU was reading 75c!! *And suddenly a CPU overheat warning on post next reset.*

I don't know what happened.
Also the corsair radiator wasn't even warm to the touch even when at 5ghz the bios said 54. Shouldn't it have been warm(took fan off n touched it)

So as of now I ordered all 3 parts again to rule out parts. Even the board. Not sure if I should wait for new board or try CPU n new cooler on old one. Still debating


Here are images of the thermal:


and I did find cpu and mem ok led issues reported on the net too. Still would feel safer with a replacement board rather than just clearing cmos (if that even works).


This is what my cpu looked like after I removed the corsair yet again (previous had done the grain size single line down middle on cpu of the arctic mx4 matrix compound:
IMG_1125%20after%20removing%20plate%20had%20done%20the%20arctic%20method%20before.JPG

Then the backside of the plate looked like this:
IMG_1129%20after%20removing%20plate%20backside%20looked%20like%20this%20after%20arctic%20method.JPG



So I used 90% alcohol and started over again..

This time I used the “credit card” method I’d read about online, made a nice thin coating just on the cpu like this:

IMG_1131%20next%20attempt%20this%20time%20with%20credit%20card%20method.JPG


That should have been perfect or so I thought.. I was still reading 55C in the bios at 5ghz (I felt this reading should have been more like 47-50 max)..

So this (what should be a good paste this time) coupled with the fact the radiator wasn’t even warm, led me to believe that the corsair h70 pump must have an air pocket or be defective.

Then after a couple of resets and trying to tweak the clamps on the h70.. is when the craziness erupted with red leds etc.

Does the last paste look ok? I thought it was thin enough.
 
Last edited:
Less is more, a lot less.

Metal on metal is best for heat transfer. The thermal compound is suppose to fill in the gaps in the surface of the metals.

The credit card method is usually for exposed dies. Stick with the thin line across the cores and do not spread. Have the heatsink flatten the compound when the heatsink is mounted.
 
Last edited:
Less is more, a lot less.

Metal on metal is best for heat transfer. The thermal compound is suppose to fill in the gaps in the surface of the metals.

The credit card method is usually for exposed dies. Stick with the thin line across the cores and do not spread. Have the heatsink flatten the compound when the heatsink is mounted.



So the way i had it in my first image was probably better (the end result after i had done the thin line, but maybe too much even then? it also didnt cover the whole cpu as you can see).
 
Its ok that it doesn't cover the whole cpu. The actual die of the cpu is under that metal plate on the cpu and near the middle, so the corners aren't important.

http://www.arcticsilver.com/pdf/appmeth/int/vl/intel_app_method_vertical_line_v1.1.pdf

Also what the temps are in the bios doesn't mean much. Report your load temps in Windows. That said if your getting really hot temps to set off a warning in the bios that obviously isn't good.
 
Last edited:
Just to be sure we are talking about the thin line method right? Notice the coverage area isn't the entire cpu. Just a radius around the center of the cpu.

Yep the thin line method..

Maybe my line wasnt thin enough.

Or, maybe my technique to remove the old was flawed.. i did use 91% isop. alcohol though.
 
Some more info..

I can now boot.. but.. i cant put more than 1 stick of memory in, without error code 38.. and memok red led.

I'm reset the cmos..

I'm not sure what else to do..i'm at the latest bios 1204

I cant get a replacement board until at least april which sucks.


I'm in windows now with only 4GB instead of 16GB.. trying 4.8ghz again.. testing prime95 blend.. after 15 minutes, roughly 71C max so far.. i guess having all that extra memory can spike the temp.. as by now at 5ghz w/ 16gb i'd be about 79C. update: after 18 minutes with only 4gb and 4.8ghz, up to 79C, so i guess its all the same, oh now at 81C, 80C etc

I suspect my board is bad though, since it wont take more than one stick
 
I'd recommend Prime95 Small FFTs for testing cpu stability and heat figures. I usually play an 1080p HD video as well just to increase the load on the PC. Can be good entertainment too, while you wait. My processor runs 4-5c less with just the blend test.


As for the quirks with the board, I had my share as well. My Asus board has done some weird things. After I reset my bios, my board wouldn't post. I tried a bunch of things like leaving it unplugged for 15 minutes, leaving it unplugged with the bios jumper in the clear position, etc. Finally something fixed it. When building it, I also had problems with it being finicky with the Sata configuration I was using and a PCIe card that wasn't fully plugged it when I screwed it in. Lots of things caused it not to post.

Basically I'd say if you have time play around more, you might find your ram slot are not dead, but the board is being finicky about something.
 
Yeah def prime 95 64. And make sure you guys enable HPET 64 bit mode in the bios. Prime stresses the cpu more than linx or ibt on this cpu. Wierd as hell. EDIT - actually, DISABLE HPET 3264 bit mode.
 
Last edited:
I too have the P8P67 Deluxe and had terrible trouble with the latest BIOS....these boards are niggly at the moment. I went back a rev. of the bios to 1053 and the board is much better overall.
 
maybe ill try going back to 1053.. i have a replacement cpu coming tomorrow and another h70.. i'm kinda hoping they will fix my ram slot issue magically lol

I had my work shop check the traces on the board, where i thought i might have done some harm.. at least with what we tested, it appears the traces are fine

From what i've read, 67-70 max should be what i'm seeing at 4.8 on prime95
 
Last edited:
Well, new cpu, tried again..

Board still wont post with more than one memory stick.. did the cmos clear again..

The error code on the led is 45 or 4S, the manual doesnt have this code listed.

Its looking like the board is at fault :(

i will try a different bios soon though, or maybe psu if i can find one.
 
Confirmed, board is bad.. swapped video card, swapped psu, swapped ram.. all the same error code 45. Put on bios 1305 too.

Now i'm faced with a big question, do i keep the cpu that hit 5ghz (though since the board went bad, hasnt been stable there).. or keep the newer cpu, with the chance i cant make it stable (or at least i hope for 4.6 minimally).

Ill know tomorrow when the replacement cooler arrives, though, if the cooler was bad.. i can boot with 4GB of memory, but cant survive even at 4.8ghz in prime95.

I'm even thinking maybe i shouldnt run this thing over the next 45 days till i can get the replacement board, as maybe whatever is wrong could affect the (new) cpu or the ram etc.
 
Back
Top