Sign in with
Sign up | Sign in
Your question

Q6600 - 2 good cores, 2 bad?

Last response: in Overclocking
Share
August 14, 2007 3:22:24 PM

OC'd to 3Ghz (333) @ 1.265 Vcore.

I set up 2 instances of Orthos with affinities set:

1st Orthos - Cores 0,1
2nd Orthos - Cores 2,3.

Cores 0 & 1 always run hotter than 2 & 3. After 4.5 hours, the Orthos on cores 2 & 3 stopped while the Orthos on cores 0 & 1 (the hotter ones) ran for 8+ hours until I stopped it manually.

Does that makes sense? the cooler cores crapped out first? Does hotter/cooler not really matter and I just need to add a slight bump to voltage?

Thanks,

More about : q6600 good cores bad

August 14, 2007 7:55:32 PM

A difference is normal - I have never had more than 2, usually one, cores fail in orthos. Adjust vcore and continue unless it is already too high, in which case you should reduce fsb.
1.265 is probably in the low side.
August 14, 2007 8:08:32 PM

Do you have a G0 Q6600 or B3 stepping? If that thing is B3 then you probably want to increase the VCore... I really don't think 1.265v is going to cut it.
Related resources
August 14, 2007 8:26:36 PM

It's a G0 from ZipZoomFly (packed by Intel on July 27th) for $285 if anyone's shopping...

I only ran two instances of Orthos. The first was set to cores 0 & 1 and the second was set to cores 2 & 3.

Should I have 4 instances for a Quad? One instance per core? I'm sure what happened was that one (from the pair) failed and screwed up Orthos for the other. What's wierd is that the cooler core(s) failed, not the hotter ones.

My goal for OCing is 3.2Ghz but not above 1.35V Whatever 1.35V gets me I'll be happy with.
August 17, 2007 1:12:25 PM

UPDATE:

Is it possible to have 1 bad core that always fails? My Core # 2 is 90% the core that fails stress testing in Orthos. Occasionally Core # 0 will be the one, but Core 2 is my bane for OCing.
August 17, 2007 10:17:27 PM

It's possible. Try increasing your voltage to 1.35 then back it down until you find the min.
August 18, 2007 12:07:39 AM

I have sort of had a similar issue. When first tinkering with OCing my Q6600, my core 0 and 1 orthos was the one to ALWAYS fail. I have not had an OC and voltage yet that core 2 and 3 have failed. at 3.2 ghz with 1.30 vcore cores 2 and 3 ran fine while cores 0 and 1 would only run orthos for 2-3 seconds before failing. Quite odd. Currently I am at 3.0 @ 1.318vcore, so your definitely in a better OCing boat then I am.

So basically, I'm not sure if we both have a bad core or it is normal lol.
August 18, 2007 7:58:25 PM

actually, stress testing for 16+ hours revealed that I needed to up the voltage. I want 24 hours solid before I'm satisfied. I'll post my results this week.
August 19, 2007 10:17:18 AM

You should consider lapping your chip and HS if you wanna keep it; it will reduce the temp differences between cores. Also, as others have suggested, a high vcore is likely needed. I might consider RMA'ing your chip if I were you. My B3 does 9x333 @ 1.2625V in the BIOS (1.232 V in CPU-Z) and is stable to p95. A G0 should be well under that in my opinion.

Have a look at this thread if you haven't already which shows the mean VID for G0's and B3's.

What is your VID and is it under the mean for G0's?
August 20, 2007 8:21:23 PM

Do you use two instances of Orthos for your testing?

Also, if I RMA, what's my excuse and do I exchange or return?

never RMA'd a CPU before.

Thanks,
August 21, 2007 1:15:09 AM

Hey everyone. After trial-and-error, I've found my problem.

BAD RAM

After several tests and ram configurations, I've found that at least one of my 4 sticks is bad. Always causing errors while the others didn't.

I'm back down to 1.265v to see where I go from there. I hate starting over!
August 22, 2007 1:11:19 AM

Hmm that is interesting indeed... I need to run memtest then. I was just running an Orthos test and had core 0 and 1 fail on me after 1h 24m with 333x9 @1.318 ... perhaps I'm having a memory issue too... I suppose I will have to test this out.
August 22, 2007 3:20:47 AM

I didn't have the patience for Memtest. I cranked my FSB up to 350 (3.15Ghz) and my voltage to 1.3v and let Orthos Run till it crashed. RAM is Cas4 Corsair DDR675. I loosened the timings to 5-5-5-15 @ 1.9v to run these tests.

KEY:
2GBa - 1st pair of RAM
2GBb - 2nd pair of RAM
2GBc - 3rd pair or RAM (DDR800)

Below are my results (times are rounded):

TOGETHER
2GBa + 2GBb = ERROR @ 9 min
2GBa + 2GBb = ERROR @ 10 min

ALONE
2GBa = 15min CLEAN
2GBb = ERROR @ 4 min
2GBa = 15min CLEAN
2GBb = ERROR @ 7 min


And then I tried some PC6400 2GB kit

2GBc = 15 min CLEAN

So, without the need for Memtest, It's blatently obvious that my second set of DDR675 2GB RAM is the culprit. I really did go back and forth to make sure the errors weren't flukes. So far, I'm 14 hours stable and counting on Prime95 v25.4 (loading all four cores) @ 1.265v. Let's see where it takes me....
August 22, 2007 4:01:30 AM

I'm kinda relearning about overclocking ...haven't done in many years.
I figured I'd post my info in case it helps give a few more data points for the overclockers out there.

Here is what I have so far:

Q6600 GO
Abit IP35 Pro
Tuniq Tower (lapped)
Corsair XMS2 2Gb 4-4-4-12
Antec 900 case = wind tunnel
Vista 32 Bit

I've got it to 3 GHz....9X334 with 1.240 VCore

I'm running Prime95 and its been running nicely for 3 hrs now

Temps: Cores 0, 2 = 50 C
Cores 1, 3 = 61 C

Its pretty cool in my room too....18 C

I'm surprised that I'm getting almost 10 C more on 2 cores. I lapped the heatsink too...thoughts? I don't want to lap the processor.

Also, what version of Orthos you folks using? The version I downloaded doesn't seem to want to work with Vista 32bit. Is Orthos necessary to stress test or is Prime95 sufficient?


Thanks
August 22, 2007 4:09:55 AM

If you're using Prime95, make sure it's v25.4. It'll run all four cores 100% simultaneously. According to graysky (who wrote the dual/quad OC guide) he prefers Prime95 over Orthos due to ease of use. I can't get Orthos to run on all four cores unless I have 4GB of DDR anyway.

Once I get 24hrs stable @ 3Ghz, I'm gonna shoot for whatever I can get @ 1.35v or less!
August 22, 2007 4:17:36 AM

Graysky....nice OCing thread. It's gotten me this far.

This is kinda bizarre.... I reported to you that my VID was 1.1625. I was psyched cause it was so low. Well, now as I look at Core Temp 0.95, it reports the VID as 1.2750. Does that make any sense? Does the VID change with overclocking? I thought it was a tested value of the processor itself and wouldn't change. Any insight?
August 22, 2007 4:21:07 AM

krazynutz said:
If you're using Prime95, make sure it's v25.4. It'll run all four cores 100% simultaneously. According to graysky (who wrote the dual/quad OC guide) he prefers Prime95 over Orthos due to ease of use. I can't get Orthos to run on all four cores unless I have 4GB of DDR anyway.

Once I get 24hrs stable @ 3Ghz, I'm gonna shoot for whatever I can get @ 1.35v or less!



Yeah...that's the version of Prime95 I'm using.
August 22, 2007 4:30:00 AM

wow. if you can get 3Ghz @ 1.24, that's pretty sweet. Keep us posted! Do you have a Ghz goal?
August 22, 2007 4:35:50 AM

A couple more tidbits of info...I found the screenshots I did when I initially booted up.

My stock temps about 5 hours after 1st boot of my newly built system:

27-28 degrees C on all 4 cores (No OC, No load). Not much variation in temps at all (not pushing them I guess). The VCore at these temps was 1.050 as read by CPU-Z.

And I'm not crazy...the screenshot clearly shows my VID as 1.1625....go figure?
August 22, 2007 4:45:05 AM

krazynutz said:
wow. if you can get 3Ghz @ 1.24, that's pretty sweet. Keep us posted! Do you have a Ghz goal?



Honestly ...I'd like to just be at 3 GHz with nice cool temps and stable. A 25% OC isn't too shabby for an old goat. Hell... the last system I OC'd was a 1.0 Gig Athlon many moons ago. But I'm gettin the bug and I think I might just try to see where this thing takes me :D  I'd love to get it stable at 3.2 or 3.4, but I'm pretty sure from what I've read that the voltages to do it will bring my temps very high and I don't want the CPU throttling itself down. That's a bit too scarry for me. I'd like this processor to live a long life. I might try for my high OC and then turn it back down for everyday Ops.

I do want to try to do something about the 10C difference in core temps though. Don't know if I'll need to remove the HS and clean, then reapply the goop...we'll see
August 22, 2007 4:49:59 AM

well, I have a 7º temp difference between my hottest and coolest cores at load. They're all within 3-4º at idle. Temps, currently while running Prime95, in order (0-3) 59, 56, 52, 54

Hope that helps.
August 22, 2007 4:58:00 AM

Thanks for the info...what voltage you at now for those temps?
August 22, 2007 2:02:04 PM

DOH!

14.5 hours and ERROR @ 1.265v (which those temps are at)

Right after I went to bed! Frustrating....

Upping to 1.285v while I'm at work (fingers crossed)
August 22, 2007 3:37:23 PM

3.0 GHz .....12 hrs Prime95 stable - no errors and my temps are rock steady 50, 60, 50, 60

1.250v in bios, 1.24v shown in CPU-Z

I will never work that CPU that hard ...ever....so I'm calling it good for now.

I might check to see how low I can put the volts and have it stable at 3.0 Ghz...minimize the temps ...sweeeeet!

or ..... maybe I'll try for 3.2 next...
August 22, 2007 3:42:18 PM

Yeah, I have to get that 24hr mark because I do 3D work and if I set up a render, it might go over a couple days (if I'm doing animation) so I need that security.

Go for 3.2Ghz 12-hour stable. I'm curious to see what voltages you need for that.
August 22, 2007 3:54:58 PM

krazynutz said:
Yeah, I have to get that 24hr mark because I do 3D work and if I set up a render, it might go over a couple days (if I'm doing animation) so I need that security.

Go for 3.2Ghz 12-hour stable. I'm curious to see what voltages you need for that.





I'll keep you posted
August 22, 2007 4:18:12 PM

I'll try myself once I get 24 stable at 3Ghz. I'm at 1.285v right now. If I have to go up to 1.325v to get 3.2Ghz, I doubt it'll be worth it. Now, if I can get reasonable temps @ 3.4Ghz @ 1.325v....that's another story

And wishful thinking....ha ha...
August 22, 2007 4:21:59 PM

krazynutz said:
I'll try myself once I get 24 stable at 3Ghz. I'm at 1.285v right now. If I have to go up to 1.325v to get 3.2Ghz, I doubt it'll be worth it. Now, if I can get reasonable temps @ 3.4Ghz @ 1.325v....that's another story

And wishful thinking....ha ha...



Good Luck :sol: 
August 22, 2007 8:57:02 PM

Well I'm a little concerned with my vcore and performance. I'm thinking I may have a memory issue like you had. I ALWAYS have core 0 and 1 fail on me. I had my vcore set at 1.312 and had an error after 1h 24m at 3.0. Then I pumped it up to 1.318 and had it run for 10h 24m before I had an error on creo 0 and 1, while 2 and 3 were chugging away. It almost seems like 2 and 3 would run stable at 3.0ghz at 1.28v while 0 and 1 die instantly. I don't know if I have a memory problem or a bad cpu or what. But even with that high vcore I still have 59,57,52,52 on my temps under orthos, but my room is pretty cold.

Another odd thing I've noticed is that core 0 is ALWAYS the hottests, and then, randomly one other core will be almost the same temperature, and the other 2 will be 5-9 degrees cooler. Its odd. I also noticed that core 0 and 1 go through the orthos tests slower then 2 and 3, but I'm wondering if backround programs are running on 0 and 1 more which will slow it down.

I suppose I'm just gonna have to run memtest to figure out if my memory is good or not.
August 22, 2007 10:39:32 PM

abrahm, is your memory fairly new? Definitely run Memtest @ 3Ghz but only use ONE memory stick at a time. You could run Orthos the same way. Yes, you'll only have 1GB RAM in single-channel mode, but you're not looking for performance right now, you're looking to isolate the problem. Keep us posted on your results. I didn't bother with Memtest because, as you see in an earlier post, it was obviously my second set of RAM. Good luck!
August 23, 2007 4:33:21 AM

Krazy, yes the sticks are new, about a month old. I'm running memtest as we speak but I already started running it on both sticks before I read this. I guess I'll see if I get any errors, and if I do I'll start pulling sticks. If I do get errors, I suppose I'll order another set of this ram I have(2x1gb g.skill pc6400) and once that gets here, I'll rma the ones I have, since I wanted 4gb for when I update to vista and I'll need ram during the rma process. I'll definitely keep you posted.
August 23, 2007 3:50:50 PM

Well I just stopped my memtest after 31 cycles through, with 0 errors. I'm wondering if something else could be causing my instability, such as my psu. Its an enermax 485, which is a good psu but may be a little stressed out for this new build(blows slightly warm air out at idle and some warmer air out under loads). Either way I ordered a Corsair 620HX so whenever I get that I'll throw it in and retest. But, after reading the G0 overclocking thread on Overclocking.net, I don't feel as bad about my clock. I'm seeing a lot of people over there running 3.2ghz, 3.3ghz on 1.4vcore. That .08 volts higher then mine at 3.0ghz. I'm guessing I could atleast match that if not beat it. Also, there are guys running 3.6 at 1.5v, which seems rediculous to me lol. Oh well, I'm happy with where I'm at now, almost 12 hours of Orthos seems good to me since A) My computer will never be stressed THAT hard, and B) Even if it did somehow get stressed that hard, it would never be stressed that hard for that long. And with a better cooling solution(using Arctic Freezer 7 Pro, good but not the best) I'm sure I could bring my temps down even more. I'm not so worried about my temps now, as I am in a 18*-19* room and running very cool, I'm worried about when I go home for summer and my computer sits in an 85*F room. Thinking about going water around christmas. Oh well we'll see, Ill keep you posted if anything new comes up.
August 23, 2007 4:38:46 PM

From my readings, I've seen people go from 1.25v to 1.325v to get 3Ghz 24hr stable. That's a pretty big gray area for the G0's. I guess it means that with the lower voltage chips, they're very different from each other as far as OC voltages go. Very interesting. I should be stable @ 1.285v for 24hrs of Prime95 v25.4. I'm also using PC6400 Cas 4 and I might drop my multiplier down to 8 and shoot for 400Mhz (3.2Ghz) to run my ram at stock speeds. I'll be more than happy with that as long as the voltages stay lower than 1.35v. I'll keep you posted with my results as well.
!