65 nm quad fun!!

[Sailindawg]

Working on my clocks and so far I'm doing alright at 9x437. Not real extreme, but it seems pretty solid so far. I can run 20 runs of ITB, memory is stable with memtest, 3D game stable & 3d bench stable. I just gamed for a couple hours, no crashes or wierdness. All on about 1.35 Vcore, 1.4 NB (GTL's all +13) & 1.275V VTT. I have got to run P95, if at least 8 hours stable, it's a go! Nice clocks without having to use very high volts!

What's surprising me is the OCZ RAM. It's $50 1333 Gold series RAM, rated at 9,9,9,27 at stock. I'm running it 6,6,6,18 PL6, CR1N on 1.84 Vdimm. No extra cooling needed. The RAM is very cool to touch.

I've set 1.3675 Vcore in the BIOS, the BIOS reads 1.35V and Aegis reads 1.35V in Windows. Vdroop compensation is enabled. Under load, Vcore drops to 1.34V. Not earth shattering clocks, but it seems to be running well so far, now on to P95!!

[Chris_redfield]

Bloody hell!

Thats increidble with that RAM. My OCZ 1800MHz seems to have just deteriorated and caused me nothing but problems.

[HDCHOPPER]

ooo man exellent ram speed set respectable clock ! now for magic 4.0

[saaya]

nice! congrats

[jcool]

Nice, let's see if you can get 4Ghz.

[Sailindawg]

Bahhh.........at the original voltage I posted about failed P-95 blend after 2.5 hours on core #3. I grudgingly upped Vcore to 1.375V and P-95 blend ran for.......7 hours 48 minutes.......before core#3 failed That was this morning. Since then, I have regrouped and will continue my quest!

I believe that the rule of thumb regarding P-95 blend, is that if it fails within 2 hours, it's cpu related; if it fails after an extended period >2 hours, it could be RAM related. I backed command rate to 2N from 1N and I'll prime again. If I have the same failure, I'll up v-core another notch. My problem is probably not enough Vcore.

However, this RAM will pass all 8 tests of Memtest overnight for about 25 passes, but throw out just one error in a single Test 4 pass when other settings have been used. 1:1, this ram will post at 14xx fsb, but give me that one error as described. So far, using either the 1.3.2 or 1.3 dividers, the RAM seems well behaved. I need to rerun my ram at these 1N & 2N timings just to make sure it's solid. I have solved this before by either lowering CR to 2N or raising Vcore. If all else fails, more vcore!

[Sailindawg]

I nailed it down! I ran at RAM at CR2N, different core failed after ~8 hours. RAM was not the culprit, back to CR1N and about 1-2% bandwidth recovered. Instead of raising Vcore, I left it at 1.375. I raised VTT to 1.35. Taking 2/3's of that value, I figured that I needed 0.9V for the GTL's . Using Saaya's GTL voltage chart, I figured I'd start by using a range of values. Using 1.35V VTT, my values to use were +9 - +12. +9 being 0.91V. Reset the BIOS, 1.35VTT, +9 GTL and went back to P-95.

This time, 10 hours 58 minutes P-95 Blend stable

Now to see where else this chip can go!

[koda]

Well done
3600 seems to be hard but must be your next target

Keep pushing it !

[HawaiianSupermn]

Keep pushing bro!

[Sailindawg]

3600 seems to be hard but must be your next target

I was thinking of another direction, lowering the multi to 6 or 7 and seeing of I could hit 500 fsb. HOWEVER, that could be a waste, other than seeing what max fsb could be. I'm still only air cooling the NB with the stock fan. At 50%, the fan is quiet, however, during stress testing, I run it at 85%. At 85%, it sounds like a Hoover vacuum cleaner. To really push this board, I will need to water cool the NB. I'm liking that I can keep Vcore down so far. I have a very nice 24/7 clock with 437x9. I'm going to go back and try to fine tune this clocks with the 1.3.2 mem divider again to get more bandwidth from my RAM. Then this clocks will be set!

[jcool]

FSB 500 stable with a Q6600 is next to impossible. I wouldn't waste my time, plus it won't help performance one bit, seeing as you are way below the 4Ghz.

[Sailindawg]

I agree. I was thinking of trying it as an exercise to what my max fsb could be. Agreed 100%. I have also purposely kept voltages down to see what I could get at lower voltages first. I need to water cool the NB.

[Sailindawg]

Whelp, been going back at clocking this cpu. Here's my latest:



Vcore=1.475 V (x1 V multi)
Vdroop=enabled
PLL=1.508861 V (default)
VTT=1.425 V
GTL's=+12 (all)
NB=1.5 V
NBGTL=default
SB=1.5 V
RAM=1.84 V (running on 1:3 divider)

This OCZ Gold is still holding strong at 1353 FSB at the 66618 PL6 timings. Kind of wondering if I ought to get an 2x1G kit and go VISTA 64 as well. Not sure at all if these 1333 Gold's are the norm or I just got real lucky. The RAM seems happiest on either the 1:3 or 1:3.2 divider.

At these settings, I can also hit 9x412. So far I'm 4 hours P-95 stable, 3D gaming stable and WPrime stable. Unfortunately no screenshots. However, if I'm 8 hours P-95 stable at 8x450, I'm sure I'm also as stable at 9x412, considering I'm using the same settings. Ultimately, I need to run P-95 to figure it all out.

Overall, this cpu seems to be scaling with voltage. Temps at 8x450 ranged from 45-55 C under P-95 Blend. 55 C was hit on occasion, with 48-50 more the average for all 4 cores. I was running my rad fans on 12V to help keep temps down. Still air cooling the NB. Running the stock fan on 85% duty cycle, NB temps stayed ~45-47 C. The machine sounds like a hoover at this point! I can run things much quieter under 24/7 use. The stress tests really hammer temps.

[Catz]

Ok not trying to tell you that your overclocking is easy but 3.6ghz on a Q6600 G0.... that's like a walk in the park....

3.8 is actually easy for me aswell...

now 4.0ghz on air would be impressive.

So either i don't understand this thread or you should try harder :P

[Sailindawg]

Thanks for the encouragement! I'm a lazy overclocker taking my time to see where I can go with this board and chip. I well understand that 45 nm is the way to go, but that's not what I've got. This is my first 775 board and I'm enjoying what it can do. I suppose that I could just up the volts and see where I land, then work backwards to see what's stable at the min volts. I guess the point of the thread is that I'm enjoying the board.

[HDCHOPPER]

thats what it's all about
not every thread has to be "I got a problum" thread

get it to 4.0 and go upstairs

good steady work man keep it up my first hitting 4.0 on a pent "D" was quite the rush

[Catz]

alright sweet, i wish you all the luck

Didn't quite understand the thread

4.2ghz is my max on the Q6600! Have fun

[Sailindawg]

@Rem

No problem dah-lin' .........I'm chasing 3.8 at the moment ......at 1.6V on the northbridge, I'm getting to the limits of my NB cooling. I'm hitting 50C on the NB. Not bad, but from what I've read >50C not so good! I'll give 4.0 a go sooner or later!

[NCspecV81]

Working on my clocks and so far I'm doing alright at 9x437. Not real extreme, but it seems pretty solid so far. I can run 20 runs of ITB, memory is stable with memtest, 3D game stable & 3d bench stable. I just gamed for a couple hours, no crashes or wierdness. All on about 1.35 Vcore, 1.4 NB (GTL's all +13) & 1.275V VTT. I have got to run P95, if at least 8 hours stable, it's a go! Nice clocks without having to use very high volts!

What's surprising me is the OCZ RAM. It's $50 1333 Gold series RAM, rated at 9,9,9,27 at stock. I'm running it 6,6,6,18 PL6, CR1N on 1.84 Vdimm. No extra cooling needed. The RAM is very cool to touch.

I've set 1.3675 Vcore in the BIOS, the BIOS reads 1.35V and Aegis reads 1.35V in Windows. Vdroop compensation is enabled. Under load, Vcore drops to 1.34V. Not earth shattering clocks, but it seems to be running well so far, now on to P95!!

Your ram sounds like the Samsung HCH9's.. This board loves this ram! I've taken the 1333's upto 2.1ghz! @ 8-8-8-18 1T. PL8

[Sailindawg]

I've taken the 1333's upto 2.1ghz! @ 8-8-8-18 1T. PL8

What were your settings? What Vdimm did you use? If I use any other divider than 1:3 or 1:3.2, the board will not post.

[Ryan12]

What were your settings? What Vdimm did you use? If I use any other divider than 1:3 or 1:3.2, the board will not post.

More volts to vdimm?

[EliTE22]

nice oc, what ur highest fsb? i cant break 469, good luck with 4GHz+

[Sailindawg]

@Ryan

More volts to vdimm?

That's the crazy thing with these RAM chips. I'm only using 1.84 Vdimm. I'm really able to run them tight at stock FSB for the RAM. If I try to run them on auto, which seems like 1:1, the board will not post. I have tried the different straps, same thing. Once I got above 400, 1:3 divider seems to work real well. The old run the ram on a divider and go for really tight timings.

nice oc, what ur highest fsb? i cant break 469, good luck with 4GHz+

So far, 450. And thanks I'm working on 4 as I type this. So far WPrime stable, now to work on ITB & P-95.


Edit:
I have been trying for the holy grail of 4G. I can boot to it, run 3DMark 2001, pass WPrime 1024 run and play some COD4 single player. But the minute I try P-95 or ITB, it reboots. The voltages are really starting to get toasty as well! I need about 1.63V to boot into windows. Maybe time to try a different BIOS.

[quake6]

I have achieved 491 fsb for super pi 1m and 460 fsb stable 32m,all done with q6600

[cadupradoo]

you have very nice results... i'm jellous... I can't go over my 378x9 ... tried playing with all voltages and didn't get stability in 3.5GHz..