Xtreme Low Q6600 G0 Temp on Air !!! (Lots of Pics)

FSB 466 & 470 * 8 PRIME STABLE Preliminary Results !!!:


So I've being trying to stabilize my rams all night and I finally found out what makes my system having difficulties getting Prime Blend stable.
You guys probably have all seen my current highest OC on air with Prime Small FFT stable screenshots but with the same settings, Prime Blend will immediately fail at Test#2 in less than two mins.


After I change NB Reference voltage from x61 to x67 and the rest of the remaining configuration left the same, I can immediately Prime Blend stable for several loops.
So now out of curiousity, I say why not try the x8 multiplier route again for maximum RAM OC using 4:5 divider. Keeping the same current CPU clock speed.


So...I was WRONG about the instability using x8 multimplier to OC the CPU w/ P5K-Dlx. Now I really gotta thank my P5K-Dlx for what it can deliver....I'm 100% SOLD !


30mins + and counting (I really have a good feeling about this cos I was having tough luck even at FSB460*8 and not even Prime-able for a couple mins); next goal: FSB470 & DDR2-1175 !

http://img505.imageshack.us/img505/3...table11qf1.jpg


Still using the same Vcore, I'm afraid I'll have to bump the Vcore by one notch.
But nevertheless, the mobo is completely happy with the Q6600 G0 at FSB470 and this is something that was not possible :)
http://img255.imageshack.us/img255/6...21175przm8.jpg

You may want to find your most stable speeds with the 9x multi. instead of the 8. I think higher FSB can slow your chipset...run some benchmarks instead of prime & see which multi performs better at ~ the same mem & core speed.

Quote:

---

Originally Posted by Icer

You may want to find your most stable speeds with the 9x multi. instead of the 8. I think higher FSB can slow your chipset...run some benchmarks instead of prime & see which multi performs better at ~ the same mem & core speed.

---

The answer is FSB468*8 = 3744mhz scores significantly better than FSB416*9 = 3744 of the same clock. (CPU Mark 5432 vs 5366 respectively)
*I'm not sure if it's also increased ram clock speed from DDR2-1040 to DDR2-1170 made the improvment* Same 4:5 divider

I don't have the 5366 data anymore, I accidentally deleted it.
Expect further improvement in 3D mark 2.0/3.0 and CPU score with better Nvidia Vista 64bit driver and a clean install Vista 64bit OS. My system is preety clogged up being my test bed for the time being.

http://service.futuremark.com/compare?3dm06=2575993

Quote:

---

Originally Posted by XtremeTiramisu

After I change NB Reference voltage from x61 to x67 and the rest of the remaining configuration left the same, I can immediately Prime Blend stable for several loops.

---

Nice job, and thanks for the tip, I'll try it on my P5K. I've been running into similar observations as you have with my board (Orthos blend OK @ 3.85, but 3.9 failing on my E6600). Were you able to find an explanation somewhere on the reference voltages in the P5K bios?

Try dual orthos, or 4 x Sp2004 :).

Here is my mild overclock (See sig for batch#):

http://www.iorealm.com/photos/pub/oc/overclock.jpg

Quote:

---

Originally Posted by chriskurn

Here is my mild overclock (See sig for batch#):

---

We've the same mobo, same Q6600 G0 CPU but yours a diff batch but not big deal. With your current given voltage @ 1.4000v Idle, your chip should be able to hit up to 3.6ghz. Keep it up :)

Very Impressed @ your results Tiramisu

what RAM would you recommend for me to get on my new system(in terms of DDR mhz).?

I already changed my mined from 680I a1 to this P5K. :)

Quote:

---

Originally Posted by Gondon

Very Impressed @ your results Tiramisu

what RAM would you recommend for me to get on my new system(in terms of DDR mhz).?

I already changed my mined from 680I a1 to this P5K. :)

---

Definetely the ram in my sig ;)
If you havn't seen the screenshot of my ram that's OC'ed to DDR2-1248 or 624mhz, you should go take a look. I think it's buried somewhere in the last couple pages

kick it up there chriskurn, at that voltage you can crush 3.2ghz

Ok, this G0 Q6600 is insane. I dropped the Vcore, raised the Mult, and the CPU runs cooler :) Prime95 Perfect, Orthos, no problems either. I think I will stay at 3.4ghz until I get a better cooler.

http://www.iorealm.com/photos/pub/oc/overclock2.jpg
http://www.iorealm.com/photos/pub/oc/overclock3.jpg
http://www.iorealm.com/photos/pub/oc/overclock4.jpg

i dont think you are stressing all four cores. get the latest prime it launches 4 threads at once.

also, use coretemp to get real temps. speedfan temps are off by 15 i think. speedfan is good to check voltage though.

Quote:

---

Originally Posted by idiotekniQues

i dont think you are stressing all four cores. get the latest prime it launches 4 threads at once.

also, use coretemp to get real temps. speedfan temps are off by 15 i think. speedfan is good to check voltage though.

---

Either way (@chriskurn), doing 80% load on 4 cores @3.4GHz with a CoolerMaster Hyper Tx (~$20 heatsink) kicks ass :up:

I am using Prime95 24.14....that was 100% on all 4 cores. For some reason the load on my cores changes as p95 runs. Any ideas?

I also need a better cooler, but I don't want the Thermalright Ultra120 Xtreme. <---Its tooo big LOL

that is because of different "FFT's" are used to excercise the cpu core...i think each FFT is run for 15 minutes before switching...you can customize it to a different schedule.

EDIT:...just re-read your post and i see you were referring to load not temperature...the above statement affects temperature

I found the problem. Prime95 25.3 works on all four cores LMAO
I am very impressed with the OC'ing and temps that you get from the G0 stepping.

http://www.iorealm.com/photos/pub/oc...ock3400mhz.jpg

Quote:

---

Originally Posted by chriskurn

I am using Prime95 24.14....that was 100&#37; on all 4 cores. For some reason the load on my cores changes as p95 runs. Any ideas?

I also need a better cooler, but I don't want the Thermalright Ultra120 Xtreme. <---Its tooo big LOL

---

im in the same boat as you. im waiting for this and then im going water:

http://www.xtremesystems.org/forums/...d.php?t=151451

Quote:

---

Originally Posted by Spanki

Either way (@chriskurn), doing 80% load on 4 cores @3.4GHz with a CoolerMaster Hyper Tx (~$20 heatsink) kicks ass :up:

---

i wasnt talkin smack to him, i was just advising him on getting a true 100% load on all 4 cores at once. im diggin his oc :clap:

and also to get proper temp readings to protect his gear. thats all.

Quote:

---

Originally Posted by chriskurn

I found the problem. Prime95 25.3 works on all four cores LMAO
I am very impressed with the OC'ing and temps that you get from the G0 stepping.

---

Chriskurn:

Yeah you should stay at your current OC state cos I'm a little worried for the kind of temp displayed in the screenshots you've provided above. Taking your current given voltage right now - 1.34v and 3400mhz but you're already at a similar temp as I'm running 1.44v and 3744mhz. I'm also looking at your system temp being reported as 43c shown in both speedfan 4.32 and Asus AI suite so that means your ambient temp inside your case is quite hot.

I'm assuming the ambient temp over your place in the summer time must be in the 30s. It's recommanded that you get a better cooler to unleash the G0 Quad full potential and fight the summer hell ambient:D

Next time when you show a screenshot of your load temp, please allow prime stress load to test at least 3 consecutive tests.
Looking at your time, you've only started less than 1 min and in test #1 stage, Prime Blend (less stress on the CPU)
So for you to get a better idea of the actual load temp, wait until Prime past the first couple mins; the system loads fluctuates and so is the load temp during the initial stage.

http://www.xtremesystems.org/forums/...d.php?t=151451

when this comes out im pickin it up and runnin 3.6ghz 24/7 :D

Quote:

---

Originally Posted by idiotekniQues

i wasnt talkin smack to him, i was just advising him on getting a true 100&#37; load on all 4 cores at once. im diggin his oc :clap:

and also to get proper temp readings to protect his gear. thats all.

---

Yep, yep... I just wanted to point out what his ~$20 heatsink is doing (same one I have). I'm digging the oc too... even if you add 15C to his SpeedFan temps (to get to the 100C Tjunction), it would make an excellent budget overclock - 24/7 Quad core @3.4Ghz with a (light-weight, no fuss push-pin mount, quiet fan) $15 heatsink! That's hard to sneeze at :up:

Quote:

---

Originally Posted by idiotekniQues

http://www.xtremesystems.org/forums/...d.php?t=151451

when this comes out im pickin it up and runnin 3.6ghz 24/7 :D

---

Please do not use that junk, you would be better off with a highend air cooler.

Tiramisu, how much air are you pushing through that Ultra 120 X? I have a XP-120 on my Opteron and a G0 from NCIX on the way (hopefully as good as yours), I'm not sure if I should try it with the XP-120 first and then go water if necessary, or order an Ultra. I would like to stay with air if possible (I won't be running 100&#37; load very often) but I hate loud fans and if reasonable temps on the Ultra 120 require a loudass fan then water it is.

Are you guys sure that Speedfan is off by 15c?

Currently my bios, speedfan, and Asus Suite differ by a few degrees.

Quote:

---

Originally Posted by chriskurn

Are you guys sure that Speedfan is off by 15c?

Currently my bios, speedfan, and Asus Suite differ by a few degrees.

---

Pretty sure, ya. See what CoreTemp 0.95 says.

Quote:

---

Originally Posted by shiznit93

Please do not use that junk, you would be better off with a highend air cooler.

---

i think ill take the words of some watercoolers in that thread over your advice.

ill wait for reviews of the stuff first - but switftech has a good reputation, and if they stay true to their quality product, the 220 version of that will easily beat an ultra 120.

if the reviews dont back swiftech up, ill pick up the mid-range petra kit and beat high end air anyways.

with my ambient temps i have no choice.

Trust me Speedfan is off by a large margin. There is no way Core temp is lower than CPU temp. Core should read higher. I would add 15C for your four cores to get something close to reality. I would also double check CPU MCP, and board temps with bios and make adjustment to match bios readings.