Need Help Stabilizing Q6600 on IP35 Pro

Everything in my sig is current except that I'm only using two DIMMS while newegg is RMA'ing my other kit.

I've got my Q6600 at 3888MHz, but it isn't stable. it seems like I might be able to stabilize this for 24/7, but i don't know what to do from here. admittedly, i got here rather haphazardly as i only benched 3d06 and kept increasing until i couldn't run 3d06 anymore. here's where i ended up:

Did you try running Orthos?

Not orthos, but I did run prime95. 3 of 4 threads fail almost instantly with small fft. and when i run 3d06 it doesn't reboot or blue-screen, it gives the windows error that the "application has stopped working... click to cancel"

i took it that these are "healthier" issues that might be fixable. thinking that restarts, freezing or BSOD are more severe issues. is this off base thinking?

try unlocking your ram so that its not overclocked. i'm sure that's not the only issue at work here, but it might help.

you cant unlock ram on intel based chipsets, the only ones that allow that are NV/ATI based.

one thing i notice is you are running way to much voltage thru the SB. all you need is ICH =1.15,ICHIO 1.55...

back the fsb down till you get 3,800 and try it then... you might be on the point to where the cpu can get stable...could be cooling or requiring way to much cpuv for a near 3.9ghz oc.

while at 3,800 you might want to up the cpuvtt to about 1.5 or higher...

Evilsizer said:

you cant unlock ram on intel based chipsets, the only ones that allow that are NV/ATI based.

one thing i notice is you are running way to much voltage thru the SB. all you need is ICH =1.15,ICHIO 1.55...

back the fsb down till you get 3,800 and try it then... you might be on the point to where the cpu can get stable...could be cooling or requiring way to much cpuv for a near 3.9ghz oc.

while at 3,800 you might want to up the cpuvtt to about 1.5 or higher...

Click to expand...

thanks evil.. i wanted to try those suggestions yesterday after work, but i'm in the middle of 24hr memtest86 runs

should be done when i get home tonight and i'll try it.

on another note, I just made a deal with someone at XS for a Q6600 from the L737B batch which seems to be a superior batch. he was running 12hr prime stable at 3.8GHz with 1.3875 vCore... (maximus formula and 8500 tracers were other parts). so, i figure if I can't get that chip stable, then it's me, not the hardware holding me back lol

we'll see soon enough

btw - what is "loadline calibration"? is that a maximus thing? is that a key piece to achieving similar results? why didn't i ask this question before i paid him?! lol

never heard of loadline calibration, it might be something for the maximus...

I don't know where you got the idea that you cannot unlock the ram...my ram is running 2:3 atm, and I have many options in bios to change it...although I don't know if it can be underclocked, say 3:2 instead of 2:3..I'll have to look n see.
So far, I've had no reason to change it as it is running flawlesly

Uh, I seem to remember that nvidia boards will allow an "unlinked" state which will let you underclock the mem...so Evilsizer I gues I made your point even stronger, so nevermind the above stupidity.

Leftheaded, I seem to remember someone posting the mem ratios for intel based somewhere in this section...but I suffer from C.R.S. so You will need to search it out...there are many settings that might be just what you need to get a stable O'C.
If you look at the SPD data sheet in CPU-Z, you should see what the mem is rated at, and what it's capable of...I'll try to post a pic to show you what I mean.
Doc

thanks for the suggestions.

well, i ran 24+ hours of memtest on my two sticks of RAM and one of them had 3 errors during test 6 of pass 25 of 47. the other stick went over 100 passes with no errors. I'm going to have to RMA this kit... I'm not going to bother trying the other stick (that had 3 errors) at 2.4GHz

i should have done more memtest passes before I RMA'd the first kit. could have saved me time & money. lesson learned - 2-3 passes of memtest is not enough!!

the good news is that newegg just sent me my other kit which I had to RMA. I ran memtest on one stick for 12 hours without any errors. and the other stick is running now (no errors yet).

but i bought a new car 1/23 and so i'm going to go play with that all weekend and might not have time to try these recommendations for a few more days

thanks again!

left is using my board i think we can get him covered!

just let us know where you get stuck at Left, congrats on the ride!
gotta ask though,what did you get?

Evilsizer said:

gotta ask though,what did you get?

Click to expand...

i got a prius. it was all about mpg.. i hate cars in general. well, not cars per se... i hate the industry and it's related industries for their underwhelming achievements in innovation.

i wanted a tesla roadster, but the sales rep said the wait time is about 16 months and it's so expensive. i really hope the tesla is a success, and that someday they make a more affordable car with the similar efficiencies.

i thought about one of those at one time... but given the cost to what i make it wont be a option for a while..though im liking the honda hybrids.. then there is that MDI air car, est $11k starting when it comes to the US..ok enough off topic..


*edit*
haha yea i looked at the telsa too.. looked at the website base price $100K

leftheaded said:

btw - what is "loadline calibration"? is that a maximus thing? is that a key piece to achieving similar results? why didn't i ask this question before i paid him?! lol

Click to expand...

"Load line calibration" is basically VDroop compensation. It may or may not help with stability - load volts will be higher, but changes in load will cause the voltage to bounce around more. It's an Asus thing.

As far as the little I've learned in E.E. goes, a Vdroop mod works by changing the power transistor feedback path such that it changes the loading characteristics of the circuit (i.e. the "load line"). "Load line calibration" is, in other words, a "soft" Vdroop mod.

Vdroop mods have caveats, for example, the transient response of the circuit will be worse - the voltage will bounce around more when the load changes (more "overshoot"/"undershoot" and takes more time to settle to its new value). It's a tradeoff, worse transient response for better (less) steady state error. That's why I say it may or may not help stability - if you're benchmarking, it's always at 100% so no problem, but it might cause crashes when you start or end the benchmark.

You CAN design circuits for both better transient response and steady state error, but it's going to be $&#&ing complicated and expensive. The "pick two: cost, reliability, power" rule applies here too (TANSTAAFL).

leftheaded said:

Not orthos, but I did run prime95. 3 of 4 threads fail almost instantly with small fft. and when i run 3d06 it doesn't reboot or blue-screen, it gives the windows error that the "application has stopped working... click to cancel"

i took it that these are "healthier" issues that might be fixable. thinking that restarts, freezing or BSOD are more severe issues. is this off base thinking?

Click to expand...

Your processor isn't stable. Unstable processors can show as error pop-ups from Windows and crash to desktop.

I know, especially because Prime95 failed pronto.

I may just need to get a good water loop if I want 3.8GHz 24/7. good news is..

I got to play with my new ip35 pro & L737B Q6600 for a couple hours last night. i went right up to 3.2GHz without changing anything, except RAM volt. had to up vcore for 3.4Ghz.. i went straight for 1.4v

i booted up ran prime95 for 5min.. watched temps. all good.. then went for 3.6GHz... same thing, all good. went for 3.8GHz and one prime95 thread crashed instantly. i tweaked some other volts as recommended here, then manually set my RAM timings to 5-5-5-15-45-2T rest on auto. no luck.. tried to run 3d06 and BSOD

backed down to 3.7GHz and ran 3d06 to post my highest single-card score. but prime95 still fails one thread instantly.

my temps are 70C under load with TRUE.. i've got some learning and research to do. probably start looking at it again this weekend

for now, i've backed down to 3.6GHz and have prime95 running all night but i forgot to check it this morning! lol i'm already at work, so it'll have run 18 hours by the time i can check it.

i think 3.6GHz is 24/7 with this new L737B (unlike my other batch), and with better cooling and a little more knowledge/help i might get 3.8GHz stable 24/7. that's the goal

sorry for the double post here, but i wanted this question to stand alone from the post above...

do I need to reinstall the OS since I got a new motherboard? it's the same mobo (abit ip35 pro), but could that be part of the stability issues?

isnt 1.6v a little high? isnt around 1.4-1.5 a max for the 6600 ? and even 1.5 is pushing it for 24/7 use?

1.6v max on air i have seen some, though i wouldnt suggest that high unless on water.. thats just me..

yeah, 1.6v is high.. it was 1.57 in cpu-z, but still high. but i'm not using that anymore. new cpu batch and now i'm at 1.4v

still having stability issues... i probably should have started a new thread... anyone reading just the first few posts would be confused since i've changed settings and cpu batch. goal is the same: 3.8GHz 24/7