q6600 @3.6 - please help with fine tuning

D

dignatec

Weaksauce
Joined
Mar 25, 2008
Messages
95
Hi everyone,
I've finally overclocked my q6600 (g0) to 3.6, stable for 25 mins so far (will keep it running after I post here) but I need some help with fine tuning and have enclosed screen shots.

My specs:
q6600 (g0)
p5k premium
antec 900
TRUE hs with scythe sflex
600W OCZ stealthxtream psu
xfx 8800gt
2x1gb patriot extreme performance ddr2-800.

First issue is to reach 3.6, my vcore in the bios is very high (1.55). I've enclosed my bios settings, is there anything I should enable/disable that will allow me to lower my vcore (the vid of the chip is 1.325- fairly high)?? When I used any vcore below 1.5, prime would fail right away. My ram is running 1:1. Note the "spread spectrum" options are disabled, photo was cutoff.

IMG_1561.jpg

IMG_1569.jpg

IMG_1570-1.jpg


Here is my windows prime95 (25 mins testing so far) screen shot, as you can see the vcore under load is high 1.525, and my temps are 71 and below (will worry bout temps later).
3-625minstable-1.jpg


So the question is, are there any settings in the bios I should change that will allow a lower vcore while maintaining stability? Should I enable/disable something that I haven't done already? What should I lower first? Vcore and see what happens? Just need some fine tuning help here.
 
First up, grats on getting that moving along at 3.6 now. I would try to stretch out at least 8 hours of Prime 95 with those setting to make sure it solid there before you proceed any further, then try dropping vcore and checking again. Instead of playing around in the BIOS trying to keep tweaking it, you need to establish a good baseline stable spec and then work from there. If you keep making changes before properly checking stability, you end up chasing your own tail quite a bit and the whole process takes longer.
 
Hey Ocell,
Thanks man, it's been taking a while. Dumb question for you, is it safe to keep the vcore/temps this high for 8 hrs? I'm still new to all of this.

I'll run prime tonight while I sleep (8 hrs +) and post screenshots tomorrow. I did try to lower the vcore to 1.5 and it wasn't prime stable at all (failed at 2 mins), this is why I think I have to tweak my bios settings.
 
Well there is a certain point where you need to determine what you are comfortable with. 1.55 vcore is higher than most people would run, however its not like sticking a blow torch on the CPU or anything and killing it instantly. Also I am gonna say that I rarely gain a lot out of massively tweaking a BIOS, I normally restore the BIOS to defaults (optimized defaults if available) then adjust as few things as possible. This makes your results more somewhat more consistent, and that the fact that is heavy tweaking outside of voltage/bus/multiplier/power management (+sleep states) rarely give you a notable increase in your OC or ability to run lower voltage.

Regarding the BIOS settings, every time I go into my BIOS I restore optimized defaults, then adjust what I need to. With my current Q6600 + Giga DS3L, that's turning off on board sound, turning off fan power management, lowering the multiplier to 8, bus speed to 400, RAM multiplier to 2.00, disabling "turbo", and setting voltages to auto optimized. Then I adjust from that baseline (side note: I could just save these setting, however I have them burned into my brain now)

If 3600 is really your goal and you know you need 1.55v, start from there. Until you make it for an extended period of time at 1.55v no errors, consider it unstable. Also you would find that you can probably run just under 3.6 with A LOT less trouble than you are going for right now. Its cool and all to hit the 3.6 number, however sometimes its not worth the trouble. The max I could get stable at stock vcore was around 3.35 GHz, however its *barely* faster than just running at 3.2.

Right now I am stress testing my Q6600 at 8x410, stock voltage. Once it passes about 12 hours of that, I will lower it down to 8x400 so I can run at a nice even number of 3.2 GHz and not worry about it again. I can get into Windows and play around at 3.6, however I am trying to run stock voltage so when the summer hits and the AC goes out so it gets up to 90F+ in my house, I do not have to worry about the overclock.
 
Hey,
Very true points you mentioned. My cpu is very comfortable at 3.2, temps <55 C and most settings are on auto (even vcore), I simply changed the fsb to 355x9=3.2. There was no way to get to 3.6 without cranking up the settings. You're right, I don't care about 3.6 vs. 3.5, this is just for fun right now. I'll see if I can hit 3.6 stable, and then drop down to 3.5 and start lowing all the voltages, hopefully 3.5 requires a lot less voltages and I'm totally happy with that.

I'll post back in the AM if it's stable at 3.6 for 8 hrs. Also, is it normal for my temps for core 0,1 to be like 67 , while for 2 & 3 to be 58? It's about 9 C lower for the 2nd pair of cores, which makes me think the heatsink wasn't installed correctly by the computer shop.
 
dignatec said:
I'll post back in the AM if it's stable at 3.6 for 8 hrs. Also, is it normal for my temps for core 0,1 to be like 67 , while for 2 & 3 to be 58? It's about 9 C lower for the 2nd pair of cores, which makes me think the heatsink wasn't installed correctly by the computer shop.
Click to expand...

I had a similar delta between the core temps until I reseated my CPU cooler. Now my temps do not vary by more than ~4C.
 
Hey,
Ok, I'll have to look into reseating. I might just take it into the shop and let them do it so I don't void the warranty.

As for 3.6: One thread failed after 1 hr, but the other 3 stayed stable for 8 hrs +. The only things I changed from above was change the NB voltage to 1.4 (from 1.55) and disabled static read control (someone suggested to do this). Might need to up the nb voltage and try again tonight.
 
If one thread fails, stop the whole thing right away... running it with only 75% loaded means it's probably only producing 80% of the total heat, so you're just tricking yourself. Odds are if the core that failed didn't, another one would have.

I had to up my VCore to 1.4875 to get my Q6600 stable at 3600MHz. I didn't change any other settings except the FSB, ram timings and ram voltage, everything else is stock.

My VCore is reading 1.39/1.40V in speedfan/cpu-z so I'm getting about .09 Vdroop.

With these settings, it peaks at 65Cx2, 58Cx2 with the side panel on and running prime95. I'd recommend getting a 120x38mm fan though for the TRUE or Ultima90 heatsinks because of the fin density. A 120x20mm fan doesn't have as much static pressure and you can lose a lot of CFM.
 
Logan: Problem is I run these stability tests at night time while I sleep (I do want to have SOME fun with the computer when i get home from work...lol). I"ll look into the fans you're recommending, the Scythe S-flex is 65 cfm, which I thought was pretty good given that it's practically silent...

Here's the problem: last night I went to test 3.5 and it was stable for 1 hr. I wake up this morning, and my tv says NO SIGNAL which tells me the computer hung sometime last night (wouldn't respond), so I'm stuck again lol. I need to wait for the weekend, so I can monitor the computer to see what's going on.

1.4875 won't let me boot into windows at all, I need a min of 1.5 vcore (maybe b/c my vid is very high 1.325).
 
dignatec said:
Logan: Problem is I run these stability tests at night time while I sleep (I do want to have SOME fun with the computer when i get home from work...lol). I"ll look into the fans you're recommending, the Scythe S-flex is 65 cfm, which I thought was pretty good given that it's practically silent...

Here's the problem: last night I went to test 3.5 and it was stable for 1 hr. I wake up this morning, and my tv says NO SIGNAL which tells me the computer hung sometime last night (wouldn't respond), so I'm stuck again lol. I need to wait for the weekend, so I can monitor the computer to see what's going on.

1.4875 won't let me boot into windows at all, I need a min of 1.5 vcore (maybe b/c my vid is very high 1.325).
Click to expand...

Monitoring the computer would not have gotten you anywhere, it probably just screen blanked with no warning. Stability checking is pass / fail anyway, staring at the screen while it happens is just a waste of time. So you do not need to know what's going on, you just need to know that test was a fail.

Currently you have no data to indicate that chip will be stable at 3.6 regardless of what you set it to. I think you should clock it DOWN to like 3.2 or 3.3, check stability (8+ hours), and then work your way up instead of hammering on the higher speeds.
 
Ocel: Actually I am currently running 24/7 @ 3.2 and it is 100% stable (14 hrs prime95) with most settings on auto. 3.2 was very easy to reach and my temps are 55 C and below under load. This is why I'm going for a higher OC.

I know staring at the screen is a waste of time lol. I just don't know why my screen blanked out as you said, I can't tell if it passed or not so I have to re-run the test again, it could have been stable for all I know...... It was 1 hr stable before I went to bed last night.

I'll retest again at 3.5, hopefuly the screen won't blank out this time (don't know why this happens).

Edit: I just googled it and the screen going blank could be some sort of power management setting from XP, I'll disable that crap and see what happens. This would make sense since this only happens when I leave my computer running all night.....Weird.
 
dignatec said:
I know staring at the screen is a waste of time lol. I just don't know why my screen blanked out as you said, I can't tell if it passed or not so I have to re-run the test again, it could have been stable for all I know...... It was 1 hr stable before I went to bed last night.
Click to expand...

The computer failed, thats all you need to know :)
 
Well if you screen blanked from a power saving mode hitting a key on the keyboard should fix that, so hopefully you tried that =P
 
Are you sure something on the board isn't overheating? When I had some difficulty OCing mine, backing off on the FSB/NB stabilized the system.

I have a msi p7n platinum sli with a G0 q6600 @ 3.6. that takes like 1.42 vcore, 1.45 fsb, 1.4nb, 1.4sb. i had to make a big jump in OC due to a FSB hole so I had started out with high voltages to try to get 3.6...but dropping the board components below 1.5 stabilized it.

seems like if the comp locks up/reboots, it's often a heat issue on the board. whereas prime failure is more likely insufficient voltage. i think you should back off and try to get up to 3.6 more gradually...in fact i don't even think the q6600 is supposed to go above 1.5v.
 
JC: I know Intel says the max is 1.5, this is just to see if I can hit 3.6/3.5 stable just for fun at this point. Unfortunately with a high vid (1.325) it requires at least 1.5 vcore to boot into Windows with my chip.

However, I will try your suggestion of lowering the fsb/nb and try again tonight. What was your vid of your chip (shown in coretemp)?

You're probably right about the heat issue. 3.2 is 100% stable, so I'm now trying 3.5 since 3.6 cores would fail no matter what I did. :)
 
JC: Thanks for the suggestion, I tried using: vcore 1.5, nb 1.4, pll 1.4, sb auto, fsb 1.4
instead now and it's been stable for 2 hrs so far! another 2 go to and it's looking good at 3.5 Ghz! :) Temps under load are 68 and below

Update: It failed after 2 hrs...lol why is this so hard? It must be this blasted VID of 1.325!
 
What did it do when it failed? My overclock was generally failing when I was clicking around in windows with prime running, or right after I closed prime and the cpu utilization was dropping it would bsod. If it's doing that, the problem is still heat or just your board having a hard time with the voltage.

If you're having prime errors but not bsods/rebooting, then a slight vcore bump might fix it, and if that doesn't fix it, try raising the FSB one and only one notch, then if it still won't work raise the NB one and only one notch. Lastly, you could try a small vcore bump again after that, but if it doesn't work then you ought to just give up on it and drop back down to an easier OC.
 
JCDenton said:
What did it do when it failed? My overclock was generally failing when I was clicking around in windows with prime running, or right after I closed prime and the cpu utilization was dropping it would bsod. If it's doing that, the problem is still heat or just your board having a hard time with the voltage.

If you're having prime errors but not bsods/rebooting, then a slight vcore bump might fix it, and if that doesn't fix it, try raising the FSB one and only one notch, then if it still won't work raise the NB one and only one notch. Lastly, you could try a small vcore bump again after that, but if it doesn't work then you ought to just give up on it and drop back down to an easier OC.
Click to expand...

During my last 3.6 test, the display said "no signal" so I couldn't get back into windows. 3.2 was very easy, so I might just live with that.


delvryboy said:
My vid is 1.325 and runs [email protected] and [email protected]
Click to expand...

delvryboy said:
max temp after 12 hours of p95 were 62C @ 3.6 and 59C @ 3.2
Click to expand...

Hey, my 3.2 is @1.37 as well. What were your other voltages (fsb/pll/sb)? you used for 3.6? I can try 1.41, but I'm pretty sure I wasn't getting into windows with that vcore (needed about 1.5 to get in for some reason) even though we have the same vid.
 
You must log in or register to reply here.
Back
Top