Q6600 overclocking help

[hat]

First time I've had a setup quite like this. The only other Intel 775 systems I've had were a celeron 430 that I BSEL modded and an E2140 I had hell getting to 3GHz.

That said I'm trying to go for the 'golden' 3.6GHz, 400x9. RAM is running at 1000, 5-5-5-15. 4x2GB. One set is 1066 and the other is 800 (though tested to run at this exact speed rock solid stable in other systems, with only 1.95v. Running 2.1v now).

CPU Vcore doesn't seem to be enough. I've had little success playing with the CPU VTT voltage, but it does seem to make a difference. Around 1.35v or so I got OCCT to last 1 hour instead of 10 minutes, but as I went higher it seemed to start going back in the other direction, failing quickly again. I have other strange voltages such as NB voltage, PLL, and SB voltage to play with, but I haven't touched yet. Motherboard is a DFI Lanparty P35.

I tried to tackle this thing on my own but I haven't been greatly successful... so now I appeal to the masters.

-ed maybe some success raising NB voltage

 

[xkm1948]

What's the temperature of your CPU? Use core temp and Prime95 to check out the temp.

I used to have a Q6600 and ran into a similar situation like yours. It just wont be stable at 400x9. It took me a while to figure out it was due to an incapable cooler. I tried all kinds of voltage combinations, increased NB voltage, set load line calibration and etc.

Check the CPU temperature first and see if it is overheating. Q6600 runs hot, very very hot. It requires high end air cooling or good liquid cooling to handle a 3.6GHz Q6600.

 

[hat]

With the h70, temps do stay within an acceptable range. During tests temps hang around 60C.

 

[Jstn7477]

Intel quads put more stress on the NB than dual cores. If your NB voltage is default, you should try raising it a bit and see what happens. My 975X board did 385 FSB with a G0 X3210 with like the second highest NB voltage in the BIOS. I wouldn't go THAT far right away, but definitely something to check out if you haven't.

 

[xkm1948]

When my q6600 wont be stable at 400x9, i tried 425x8 and 465x7. Both with the same NB voltage settings and vcore settings. I got both stable in Prime95 test for ~24hrs.

Try a higher FSB/lower multipler first. See if it is stable. If higher FSB and lower CPU OC gives you a stable OC then it means it is your Q6600 that can't get 3.6GHz


Also, when you say 60C, which software do you use? Do you mean by all four cores stay around 60C or the "CPU Temp" reading is at 60C? I suggest Coretemp. Use coretemp + intelburn for a quick stability test.

FYI my QX9650 running at 4GHz full load shows 57C in ASUS PC Probe II under "CPU Temp". Using Coretemp I can see all four cores are around 55C~57C. You need an accurate core temp to be sure that your Q6600 is cool enough.

 

[hat]

Raising the NB and CPU VTT both two notches above normal got me 2 hours before a fail. Much better than 10 minutes. This seems to be working, guess I'll continue doing this.

 

[Jstn7477]

Raising the NB and CPU VTT both two notches above normal got me 2 hours before a fail. Much better than 10 minutes. This seems to be working, guess I'll continue doing this.

What vcore are you attempting along with that? My C2E X6800 does 3.4GHz at 283*12 and it is VID modded with conductive paint to 1.4125v, but my ASRock 4CoreDual-SATAII has a lot of vdroop. I'm guessing you might need 1.45v+ vcore for 3.6GHz, although I haven't had any Conroe quads other than my X3210 that was FSB limited by the 975X chipset.

 

[hat]

1.425v in the bios. Trying more vcore now since I raised nb and cpu vtt both one more notch up and it failed in 3 minutes. Two notches up and it failed in 2 hours.

 

[hat]

I have dropped my ram speed for testing purposes. Running at 800mhz 5-5-5-15, the test failed as fast as ever. My RAM speed seems not to be the problem here. Should I try adjusting the PLL voltage?

 

[a_ump]

i know you're trying for 3.6ghz, what's the highest oc you've managed to get? some boards just aren't up to it, i had a evga 680i mobo and could only get 3ghz 24/7 stable. then went to a cheap biostar p43 and it managed 3.6ghz stable. really depends.

I always started with vcore voltage, then went on to northbridge, back and forth. If i remember i had to toy with my ram as well, either matching the fsb speed and memory speed or using a 2:1 ratio. playing around with the memory/fsb ratio made a difference. mind you its been a good 4 years since. best of luck!

 

[TheoneandonlyMrK]

i know you're trying for 3.6ghz, what's the highest oc you've managed to get? some boards just aren't up to it, i had a evga 680i mobo and could only get 3ghz 24/7 stable. then went to a cheap biostar p43 and it managed 3.6ghz stable. really depends.

I always started with vcore voltage, then went on to northbridge, back and forth. If i remember i had to toy with my ram as well, either matching the fsb speed and memory speed or using a 2:1 ratio. playing around with the memory/fsb ratio made a difference. mind you its been a good 4 years since. best of luck!

I went 400*8 with 800 mem timings but couldn't get more than 3.4 24/7 stable at 1.5 core v on a reasonable abit x38 mobo

 

[hat]

lol I haven't gotten anything stable yet. I think I have a shot at 3.4GHz at 380x9, with RAM at 950. I'll be gone aaaaaaaaaaaaalllllllllll god forsaken day long at work today, great time for a stability test. 14 hours? 16 hours? who knows but if I see it stable when I get back it will surely have been a long enough test.

 

[a_ump]

I went 400*8 with 800 mem timings but couldn't get more than 3.4 24/7 stable at 1.5 core v on a reasonable abit x38 mobo

yea i think mine took 1.41X, with a vdrop into 1.39X. on that 680I LT it took 1.36 i believe just for 3ghz. that p43 tho was a life saver. 3.4ghz at 1.32. now that i think about it, if i'd ever gotten better cooling i might have beeen able to go further.

@hat
sounds as good a plan as any. don't give up ole buddy, it took me weeks of 2-6 hr sessions of tooling around to my q66 OC just right so takes a bit of effort but once you get its just awesome

 

[sno.lcn]

Off the top of my head, I think my Q6600 settings for 500MHz FSB on P35 (Foxconn Mars) were 1.6v pll, stock SB, 1.35-1.4v NB. IIRC I always ran PLL around 1.6v-1.7v on P35 and P45.

Wow, it's hard to believe how long ago benching Kentsfield was, it seems like just yesterday

 

[hat]

Well I have achieved stability! It took 1.4v in the BIOS to run OCCT for close to 9 hours.

I'm starting to think what my problem really is here is that it will take crazy vcore to get stable at 3.6. I've been all over the place with my FSB, 380, 400, 425, and 450. The only thing that passed was 380 with all voltages at stock other than vcore. I played with NB and CPU VTT at other FSB speeds to no avail. Maybe I'll try 425x8 later... still 3.4GHz, but with a higher FSB.

Well 425x8 didn't work... test bombs very fast, even with the RAM running slower than it did at 380x9. Hm... is NB voltage, CPU VTT or some other voltage most suspect in failing at higher FSB?

 

[a_ump]

so 1.4v to get 3.6ghz? that sounds about right for avg q6600, well upper avg, some q6600 couldn't get 3.6ghz. that was just the golden standard but not everyone goes gold! lol

and yea 3.6ghz does take a lot. like i said 3.4ghz for me was just just a bump from stock 1.29(vdrop) to 1.32(vdrop), so about .3 increase for 1ghz increase in speed, but that extra 200mhz took .5 vcore. Course i was always hoping to get that platinum level 4ghz q6600 at 1.38vcore.

Congrats man, keep going

 

[hat]

No I've only got to 3.4GHz not 3.6. I'm stable at 380x9 1.4v. If I try 425x8, even with slower RAM speed, the test bombs in seconds.

 

[vega22]

what nb voltage do you have it set at?

my money is on it needing more, i would also drop the ram to a 1:1 ratio for the time being and then you can play with it more later. both need nb v.

 

[Mindweaver]

Yea, the only way I was able to get past 3.4ghz with 1.4v was moving from my P35 to a P45 board. If you can pickup a P45 cheap I'd do it, because all the research I did on my P35 told me it wasn't going to get my chip to 3.6ghz. I was using an H50 Push/pull with Scythe 120mm Slipstream 110cmf fans and my ram was corsair Dominator 1066mb and just swapping boards and I was stable at 1.4v 3.6ghz. I still have my Q6600, but it's @3Ghz right now. It's back in a P35 board, because I later put a Q9550 into the P45 it's @ 3.8ghz.

 

[Lt_JWS]

@hat,

I really think its a ram issue... try the green slots and just use two sticks of ram. Also make sure GTL+ Buffer Strength is set to weak(could be wrong on this....)

I never ran OCCT or prime I simply used BF3 64 player multiplayer for atleast 2hrs, I ran 3.8Ghz with either 1.5-1.525 in bios...

tweaking help!
http://www.anandtech.com/show/2354/9

 

[qu4k3r]

so 1.4v to get 3.6ghz? that sounds about right for avg q6600, well upper avg, some q6600 couldn't get 3.6ghz. that was just the golden standard but not everyone goes gold! lol

and yea 3.6ghz does take a lot. like i said 3.4ghz for me was just just a bump from stock 1.29(vdrop) to 1.32(vdrop), so about .3 increase for 1ghz increase in speed, but that extra 200mhz took .5 vcore. Course i was always hoping to get that platinum level 4ghz q6600 at 1.38vcore.

Congrats man, keep going

My q6600 never passed over bronze level needed 1.4v for 3ghz, 1.43 for 3.2ghz, never could reach 3.6ghz even with 1.45v, all that using a x48t-dq6 mobo. :shadedshu

 

[hat]

what nb voltage do you have it set at?

my money is on it needing more, i would also drop the ram to a 1:1 ratio for the time being and then you can play with it more later. both need nb v.

For 380x9 3.4GHz where I passed stability, all other voltages were at default levels. I tried 400x9 with some success, but I still failed. I forget what CPU voltage I was at, probably 1.5v, but I had both NB and CPU VTT raised two spots above normal, and the test failed after 2 hours. Once I tried raising them both 3 spots from normal, it failed quickly.

Yea, the only way I was able to get past 3.4ghz with 1.4v was moving from my P35 to a P45 board. If you can pickup a P45 cheap I'd do it, because all the research I did on my P35 told me it wasn't going to get my chip to 3.6ghz. I was using an H50 Push/pull with Scythe 120mm Slipstream 110cmf fans and my ram was corsair Dominator 1066mb and just swapping boards and I was stable at 1.4v 3.6ghz. I still have my Q6600, but it's @3Ghz right now. It's back in a P35 board, because I later put a Q9550 into the P45 it's @ 3.8ghz.

Not really willing to swap out a motherboard to maybe get a few hundred more MHz at best.

@hat,

I really think its a ram issue... try the green slots and just use two sticks of ram. Also make sure GTL+ Buffer Strength is set to weak(could be wrong on this....)

I never ran OCCT or prime I simply used BF3 64 player multiplayer for atleast 2hrs, I ran 3.8Ghz with either 1.5-1.525 in bios...

tweaking help!
http://www.anandtech.com/show/2354/9

Well I see our standards for stability are quite different... I suspected this from the beginning. For me if I can't pass 8 hours OCCT then it doesn't quality as stable.

I'm not too sure about my RAM throwing it off. One set is rated 1066 and the other, although rated 800, has tested stable at 1000 in previous configurations. When I tried 425x8 after passing 3.4GHz stable at 380x9, I actually had my RAM running slower. I can agree that running 4 sticks as opposed to 2 would put more stress on the system, but not that it would make my memory itself unstable, and I'm not willing to lose 4GB RAM for a few hundred more MHz on the processor. Maybe the NB needs a lot more voltage than I was giving it before.

I'll play with it some more in the coming days and see what I can get.

 

[hat]

Went back to 425x8. Raised NB volts 4 levels, test bombed. Set NB back to normal, raised CPU VTT 4 levels. So far 4 minutes in, which is looking good compared to when I raised NB volts...

 

[Mindweaver]

Not really willing to swap out a motherboard to maybe get a few hundred more MHz at best.

Yea, not really worth it. I only did because I had a cruncher that I bought a P45 board for and it really didn't have all the extra stuff my P35 had like a onboard wireless card. I was bored and really wanting to get to 3.6Ghz and swapped them out, but yea it wasn't that much better not worth buying a whole new board.

Went back to 425x8. Raised NB volts 4 levels, test bombed. Set NB back to normal, raised CPU VTT 4 levels. So far 4 minutes in, which is looking good compared to when I raised NB volts...

Nice! Good luck!

 

[a_ump]

My q6600 never passed over bronze level needed 1.4v for 3ghz, 1.43 for 3.2ghz, never could reach 3.6ghz even with 1.45v, all that using a x48t-dq6 mobo. :shadedshu

haha dam, i bet you were pissed. i know i was annoyed with my 680i Lt when i got my q6600 and couldn't even get 3.2ghz stable(3.15ghz).