My q6600 OC and a few questions

[nkresho]

Ok, here's my latest rig. I've changed a few times in the last few months. This post may help someone new at overclocking somewhat as I try to explain my methods. I also am including a chart of speeds and voltages I attained with what I'm running.

My history... I started with a Gigabyte g33 matx board, an EVGA 8800gts 320, and a e6850. I was unhappy with the oc (~3.6 on air) so I got rid of the chip and got a q6600. I also had a arctic freezer 7 pro, it didn't seem to be doing the best job, so I got the TR ultra 120 extreme. It definitely lowered my temps by 10+degrees c. I also replaced my crappy matx case with the 900, and I replaced the gts with a gtx. Just recently I got an abit IP35 pro and a raptor 74gig. I would definitely say I have the "overclocking bug".

I'm very happy with my current rig, but i'd rather be able to clock it higher.

Here's the deal. I made the following chart during the overclocking process I went through with my current rig. At each stage I used prime 95 to check stability. I ran it for 1-2 hours at each stage before progressing. I know this is not 100% stable, but it was just the means to an end, so I say "good enough". I prefer 8-12 hours of prime to make me feel safe and stable, and I did at the last stable stage, which is where I'm running it right now.

So my methods section... I picked a speed out of the air. Starting with 3.2 gigahertz (355x9). Actually, my chip easily did 3.0, prime stable, on air, without a voltage increase. I got the computer stable at 3.2 and the subsequent speeds adding only core voltage in bios until it would run without errors. if it did error, I just increased the voltage by one notch until it would run stable.

For controls, I ran my air conditioning at 60degrees F for a few hours prior to testing. It made the room a steady 21-22 degrees c. This was measured with a lacrosse technology weather station, both directly in front of the PC air intake and 4+ feet away, on my computer desk. Temps were measured using speed fan and coretemp beta. Speedfan was with a +15 degrees correction factor. Clock speeds and voltage was recorded from CPUz. My prime95 was running 4 instances of worker threads withn itself. Each core was within 4 degrees of the one furthest from it (max temp = core 0, and the min core difference was <4 degrees lower).

My memory is crucial ballistix 6400, 2 gigs running near 800mhz on each run. My timings were 5-5-5-15 with bios voltage at 2.2volts. This memory runs memtest stable at these settings as well as 4-4-4-12 @ 2.1v.

The chip I'm using is a G0 SLACR with 1.225 vcore. I got it in OEM packaging, and I mistakingly lapped it before writing down any pertinent info printed on it.

The chart pretty much speaks for itself otherwise. I checked the rig with prime at each speed.

Until I tried 400x9... This gave me all kinds of problems. I added cpu voltage up to 1.655 volts and could post and get into windows, but I couldn't get it to run prime for more than 15-20 seconds. Rather than erroring, prime just stopped and shut down my computer each time. I also increased the other voltages available in my bios by one notch each then by one notch for all, to no avail. I was at least hoping for an increase of a few seconds in prime but I didn't get it.

The chart says what everything means. I kept track of what I thought was important on each run. I increased some of the other voltages when I tried 400x9, but I was too burnt out and let down by that point to add them, sorry...

Comments and suggestions are absolutely welcome. I was really hoping for 3.6+ stable. It seems I'm having a harder time with prime stability than temperatures. I guess I could up the voltage (1.7v?) a bit more, but I get a little uncomfortable over 1.6. I didn't go higher than 1.65 because I didn't get any more prime stability between 1.525 and 1.655. I would have added more voltage if it primed a second longer when I did so, but it didn't.

I was going to water cool, but I don't think it would get me too much better. I'm not having any major temp issues as of yet.



And the excel screenshot is from my laptop, just incase any of you keen-eyed people noticed that my computer has a battery indicator icon in the tray

 

[sp33dball]

Wow! That's an awesome chart you got there, way to go. That shows some serious dedication and time.

As for overclocking, as far as I know you would have more headroom in the dual cores since they won't be putting out as much heat as a quad.

You very well could have found the highest stability point for air cooling and may want to consider water or even more extreme measures if you want to push that q6600 more. (awesome chip too, low VID).

One thing, you may not even need that much voltage into your RAM for it to be stable since you really aren't running it overclocked at all. (Maybe your next project?).

Thanks for sharing your results!

 

[nkresho]

Thanks for the compliment. It took a little time, but the organization of it may help someone who needs a starting point with their chip. This seems like a more practical way of doing it anyway. Before, I was just writing the numbers down and throwing them away after it got stable. This way I have a history of success and failure documented. Every science teacher I ever had would be proud.

I will definitely be dropping those ram volts soon. I really forgot to do that after I settled at my current clockspeed.

Maybe i'll let it warm up to ~70 and disregard the voltage to see if I can get a higher stable oc. I'll have to blast the AC again.

Thanks again.

 

[nkresho]

Here's a few screenshots from while I was in the process of doing this. I always like threads with more pics...

I got these pics at the beginning of prime testing, rather than at the end, sorry they only show a few minutes. I did get over an hour at each speed though.

Here's 3.3 ghz




Here's 3.5, for some reason coretemp is reading my multi wrong, but cpuz is correct... Wierd. Also, it was set to 355x9 in bios but reads 356x9???

 

[nkresho]

I changed my memory timings to 4-4-4-12 at 2.1 volts and started oc'ing again. Here are my newest results. I still ran it 1 hour prime stable at each step. I ran it for 8 hours at 400x9, which is where I am keeping it right now. I couldn't get it over 408x9 no matter what I tried. it still wasn't a matter of temp, it just wasn't prime stable for more than 2 minutes.

I was consitering water cooling so I could run these clocks all the time, but it just doesn't seem worth it.

And the low ambient temps (15-17 degrees c) are with the computer right next to my window AC unit. I put a piece of cardboard next to the tower to funnel the cold air into the intake. I was planning on temps above 70 but it just didn't happen.

 

[Evilsizer]

drop your mulit in the bios to get a higher fsb. ie 8x400mhz fsb would give you 3.2gh and start clocking from there. while this wont help you get 3.6ghz stable this will let you use the higher fsb to same cpu speed. depending on your ram you could run it 1:1 with tight timeing or just go loose/high mem speed.

i havent seen very many q6600's 3.6ghz stable on air. normally its the ones that require a lower cpu voltage then what your using.

nice notes too!

 

[nkresho]

Thanks. Mine was a lower vcore (1.225), but it does seem to require quite a bit of juice to get up there. Down low it was good on stock volts, like other G0's i've seen. This board seems to stray farther from the bios voltage setting as I increase the speed. At 3.20ghz it's only off by 0-0.03. At 3.67ghz it's off by 0.075-0.115. That's under the impresssion that you believe what cpu-z says.

And i'm not using the stock mb fan connections.