Q6600 Overclock(?) Issue

P

PHeeNIxx

n00b
Joined
Oct 19, 2004
Messages
52
I am attempting to overclock my Q6600 ... going for 3.2 right now (8x400). Mainboard is Gigabyte P35-DS3L. Vista Ultimate 64. RAM is operating at spec ... no overclocking there.

After some playing and slowly increasing my vcore, I seem to have something Prime95 stable, however there is suddenly a new problem I can't explain, and I'm not even sure if it is related to the overclock.

Basically the last two times I have left the system running Prime95, I have come back to find it frozen, and the display is getting no signal. I have no choice but to reboot. However, upon reboot, I check the Prime95 logs and there are no reported failures. The last time Prime95 session ran for 8h (I left it while I was at work - but I could see from the logs that it was running for 8h) but it stopped at some point and ended up in this 'frozen' state.

I have a few other voltages topped up with a +0.1 just for the sake of stability ... FSB, PCI-E, DDR. Vcore is set at 1.38750 in the bios but cpuz reports it as 1.34 idle and 1.31ish under load (the infamous vdroop i have heard so much about).

Just wondering if this is something people have seen before. If the CPU core is unstable, wouldn't prime fail? Could it be related to PCI-E? I have it set to 100MHz so it shouldn't be scaling at all with my 400 FSB. Throughout my testing so far, Prime would typically fail pretty quickly (within 2h) and it would say so. Now I have no Prime failures, just system lockup. My temps aren't outrageous ... < 65ish under full load.

Any ideas?
 
I have my Q6600 at 400x9 on Abit P35-E. Vcore set at 1.49 on my board for prime stable. I would suggest increase vcore and northbridge a little more. Do incremental increase and test for stability. Vdroop can affect stability so keep that in mind while you increase vcore. System lock up is indicative of instability, too.
 
I would bump more voltage due to the facts that some things just top out for no reason, they just have limits.

I would use Orthos if I were you, not Prime95 as Orthos stresses all cores, I don't know if Prime95 does.
 
The newest version of Prime95 seems to support multi cores.

I'm trying to get my Q6700 stable at 3.6 as well. Just finished adding fans to the northbridge and pwm heatsinks / replacing thermal paste on chipset coolers on my Abit iP35 Pro.

hv43082: What voltages aside from Vcore do you have set?
 
I used to have the same issue as you PHeeNIxx, I would leave my computer folding or even just downloading overnight and would come back to a black screen. I eventually upped the vcore to 1.45v I believe and that took care of it. I may have dropped it a bit lower after a while but I can't remember exactly. That was running at 8x400.

I just did the bolt mod on my IP35 Pro, so I'm trying for 3600 (8x450) stable. So far I need to push my vcore really high due to vdroop, running at 1.6(!) in the BIOS which drops to ~1.5 in Windows. Even with the bolt mod my PWM is through the roof, so I think I need to drop it back a bit until I get better cooling on the PWM heatsink. The Scythe Mini Kaze just isn't doing the job and I don't think I put enough thermal paste on the chips.
 
prasvt said:
The newest version of Prime95 seems to support multi cores.

I'm trying to get my Q6700 stable at 3.6 as well. Just finished adding fans to the northbridge and pwm heatsinks / replacing thermal paste on chipset coolers on my Abit iP35 Pro.

hv43082: What voltages aside from Vcore do you have set?
Click to expand...

I had to raise my vcore to 1.49 for Prime95 stability on all 4 cores. However, I was able to boot up and run apps with as low as 1.41
 
First, have you tried 9x356?
Also, try relax the timings on your RAM. I assume you're running at 1:1?
With your other voltages that you upped, was the system stable at stock? Try lowering the other voltages besides vCore, especially DDR and PCIe. There's no reason to overvolt your RAM if your running at stock.
 
same thing happened on my board when northbridge and pwm areas got too hot. put fans over both, problem was cured w/o a voltage increase.
 
My pwm temps still increase quickly but they're not getting over 80c in prime95. And honestly that seems like a worst case scenario. I did a x264 encode yesterday of a movie, and each core only hits 50% on avg, though all 4 cores are used .. so not too worried about it. Using everest or in the bios, my iP35 Pro doesn't seem to indicate NB temps. Does your iP35-E?

Aside from that, replacing the thermal compound on the nb and sb (left the pwm mosfets alone since I'm not sure how safe it is to put anything on them), seems to have helped a lot. Prime95 testing at 3.6 right now and not getting any BSODs ..just a single core failure after about 20 min...so tweaking..
 
hi there, i overclocked my q6600 g0 stepping to 3.0 ghz and heres a screenshot, mine was stable for 26hrs while running prime95 small ftt's torture test with round-off checking enabled, i hope it helps with some of your questions regarding voltages and temps.......

2002651476762267862_rs.jpg
 
prasvt said:
My pwm temps still increase quickly but they're not getting over 80c in prime95. And honestly that seems like a worst case scenario. I did a x264 encode yesterday of a movie, and each core only hits 50% on avg, though all 4 cores are used .. so not too worried about it. Using everest or in the bios, my iP35 Pro doesn't seem to indicate NB temps. Does your iP35-E?

Aside from that, replacing the thermal compound on the nb and sb (left the pwm mosfets alone since I'm not sure how safe it is to put anything on them), seems to have helped a lot. Prime95 testing at 3.6 right now and not getting any BSODs ..just a single core failure after about 20 min...so tweaking..
Click to expand...

if you use speedfan, the nb temp will show up as "TEMP"
 
Cool thanks, I'm downloading speedfan now...btw I realized that core #3 maxes out at 3.5. Anytime I run prime95 at 3.6, core #3 fails immediately. So at least for me, that's about as high as I can go (as far as I can tell). Still...I ripped / encoded some tv shows today.. x264 900kbps ps3 profile, 2pass: e6420 @ 3.2 --> 1st pass 92fps, 2nd pass 24fps. Q6700 @ 3.5 --> 1st pass 133fps, 2nd pass 65fps. The second pass uses all 4 cores to 100% ..amazing!
 
Have you adjusted your GTF ref at all? If it is just the one core, you may be able to stabilize it using a different ref. Unfortunately the abit boards seem not to respond as well to ref adjustments as some boards, but it is worth a shot. Most quads like the 67% ref.
 
No I haven't messed w/ GTL ref since I didn't really know what it was. By default, it's set at 67% in the BIOS. I guess I'll read up and see if it makes a difference - worth a shot!
 
Well changing GTL ref around in the 67 - 70 range has resulted in bsods thus far. I've tried 360 x 10 and 400 x 9 mostly. I probably need to get some faster ram (1066 probably) to try any higher fsbs.

At the moment, my big issue is with the amount of Vcore I'm supposed to pushing. I was looking at an Anandtech article where they overclocked a quad using a DFI P35 board. I had a DFI Ultra-D back when I had an X2 4600+ and I know that those boards have a hell of alot more tweaking options..but they didn't really set anything obscure that we don't already have access to w/ the Abit boards.

But if you look at their Q6600 G0 OC settings... their "recommended" Vcore settings are like 1.39 - 1.43 for 3.6 ghz. I can barely get this thing to boot with a Vcore that low! My Vdroop is like .05 from the windows measurement of Vcore (ie. 1.53 in Bios, 1.49 in Win, 1.43 under load as an example). A vcore of 1.395 in the BIOS would end up being close to 1.3V under load ..and pretty much at that point the computer would freeze.

I will say this though, changing the GTL ref does elicit a reaction from the system. I think what I'll do is try the settings that let me run prime95 with the 3rd core failing and then tweak from there. I've been using the example settings w/ no avail. The thing is, that requires me to set the Vcore to 1.54 or higher.

Anyway..I'll give it another shot and report back. All this trouble for 100mhz..lol.
 
Well no luck so far at 3.6. Either I just don't have the magic combination of voltages / GTL or 3.5 is as high as it's gonna go w/ decent temps. It's not even that big a deal ... I just figured since people were getting 3.6 with Q6600s, I'd be able to do it w/ a Q6700.

Would my PSU have anything to do with this per chance? My OCZ GameXstream 600W has 4 rails (18a x 4) ... the 12V is usually at 12.24 and 12.18 under load. +5V is at 5.13V and 3.3 is at 3.2V. hmm..
 
psu's are only a problem when their voltage regulation fluctuates too much. i haven't read anything about that psu, so i'm not sure how tight it's volts stay. i'm in a similar situation. my chip does 3.42 on 1.395 but I have to take it all the way up to 1.475 to do 3.6, it's just too little increase for the dramatic up in volts for me.
 
I'm pretty sure my PSU is stable ... I ran an oc'd e6420 at close to 1.5v for a year w/ no problems. But I agree having to increase the voltage alot for such a small gain is kind of pointless. I found another thread on overclock.net that talks about adjusting other voltages and gtl to minimize vcore at high fsb speeds..so i'll see where that goes.
 
You must log in or register to reply here.
Back
Top