O/c'ing my Q6600/analysis of temps

[graysky]

EDIT: Since I wrote the below post, I have lapped my ultra-120 x as well as lapped my Q6600. The decrease in temps have been amazing for about 5 hours of careful lapping, and $20 worth of sandpaper. The most recent temp data can be found the q6600 lapping post. I will say that after a few days of load/idle, the temps have dropped even more as the AS5 breaks in. Also, I discovered I don't need to run with a 1.310 vcore to keep the 9x333 stable. It's been stable now for >5 h on 2x orthos with a vcore of 1.3250 in the BIOS which is, according to CPU-Z and speedfan, 1.232 under load, and 1.260 @ idle.

Here is the original thread:

Finally got all the parts and the system is built.

Relevant Specs:
P5B Deluxe w/ Q6600
Ultra-120 Extreme + s-flex and AS5
2x1 gig Ballistix DDR2-800 @ 4-4-4-12

Current settings are stock (9x266 and mem is 1:1).

I have four copies of orthos running small FFT and I split the CPU affinity evenly between the core. I get 100 % CPU usages across all 4 cores. I have been running Small FFTs test on each copy of orthos for an hour now. Here are my temps, please let me know if I understand this correctly:

Speedfan: 39-45C (coolest core and hottest core); system temp is 33C
TAT: CPU0 temp is 60 and CPU1 temp is 57.

So the speedfan is giving me tcase = temp at heat spreader and values are 39-45C and anything under 50 is "good."

TAT is giving me tjunction = temp within the cores and values are 60 and 57 and anything under 65 is "good."

Am I understanding this correctly and would you consider these good temps for a stock q6600? I will try an O/C once I figure out the BIOS and which settings I need to use.

 

[JimmyMcNulty]

cool i am very interested in your oc results as i am getting a nearly identical setup very soon - btw what speed is your sflex fan

 

[graysky]

it's 1600 and is inaudible.

 

[goldragon_70]

I'm interested in seeing how this goes, please keep us updated.

 

[graysky]

9x300 = stable; again orthos x 4 for 40 minutes this time.

Speedfan: 44-53 (coolest and hottest)
TAT: 68 and 65

I can't keep 9x333 stable at stock voltage and w/ 4-4-4-12 timings.

 

[graysky]

Here is a summary of the 3 FSB numbers I ran:

[code:1:57ee279783]9x266 = 2.4 GHz

Speedfan: 39-45C (coolest core and hottest)
TAT: 60 and 57.

----------------------

9x300 = 2.7 GHz = stable

Speedfan: 44-53 (coolest and hottest)
TAT: 68 and 65

----------------------

9x333 = 3.0 GHz = stable
Speedfan: 46-56 (coolest and hottest)
TAT: 71 and 67[/code:1:57ee279783]

Also, everything is stock except the FSB (speedstep is enabled, C1E is enabled, CPU and mem voltages are stock).

 

[graysky]

As I understand it, safe temps for an e6x00 are:

tcase = temp at the thermal spreader as read by speedfan <55 C
tjunction = temp at the core as read by TAT <70 C

Anyone know what "safe" temp ranges are for a Q6600?

 

[pmr]

70ºc wont hurt. around 65ºc should be the ideal maximum to run 24/7. If you can get it lower, better.

Forget TAT and Orthos temps when stressing. You won't get near those temps during games nor movie encodings

Disable speedstep, ce1 and spread spectrum. What is your default ram voltage? And Command Rate?

 

[graysky]

Forget TAT and Orthos temps when stressing. You won't get near those temps during games nor movie encodings

Disable speedstep, ce1 and spread spectrum. What is your default ram voltage? And Command Rate?

Very true. I just did a test encode and the temps were about 5-8C under that.

Why should I disable speedstep? I love that for power savings and cooler idle.

 

[pmr]

Just because the voltage may throttle back to a level where at idle the system could be unstable when OCing.

 

[JimmyMcNulty]

greysky is it stable @333 or not?

first you wrote:

I can't keep 9x333 stable at stock voltage and w/ 4-4-4-12 timings

then:

[code:1:999d50d2af]
9x333 = 3.0 GHz = stable
[/code:1:999d50d2af]

Did you change anything to get it stable? I'm hoping to get mine up to 333 thereby making it syncronous with my 667 memory

cheers

 

[graysky]

I'm having mixed results with it actually. I have been up at 4-4-4-12 @ 9x333 for about 12 h then for no reason it rebooted. I'm @ 4-4-4-13 @ 9x333 now testing. It's certainly not a heat issue.

 

[jeepn]

Keep us up to date bro. I have all my new parts (in sig), but havent assembled it yet. You get us the OC settings on air, then Ill go from there on my water set-up. Peace :twisted:

 

[graysky]

Yeah, 9x333 @ 4-4-4-12 is perfectly stable. Coretemp was crashing my machine. I have since switched to TAT and haven't experienced a problem. I suspect I can push it beyond 9x333 since my hottest core is <65 C but I don't care really. 3.0 GHz x 4 cores is pretty darn fast

By the way, I didn't change anything in the BIOS beyond the FSB and the mem timing, voltages are all stock values as are all the stuff "they" recommend you disable like C1E, speedstep, etc.

 

[dr_kuli]

3.0 GHz x 4 cores is pretty darn fast

It is, great to get there.

By the way, I didn't change anything in the BIOS beyond the FSB and the mem timing, voltages are all stock values as are all the stuff "they" recommend you disable like C1E, speedstep, etc.

So you've some space left... :lol:

 

[jeepn]

Nice... WC w/a push on the voltage should get some nice numbers. Peace :twisted:

 

[rmc]

Hi, just finished my build last week with a Q6600.
What I found was there are certain holes that if found you can get a good overclock.

I'm at 1535 FSB, x9 on multiplier, buss-384, core-3454.
memory unlinked crucial Ballistic DDR2 8000 4-4-4-12 @1000, volts 2.2

CPU FSB voltage raised to 1.5. CPU core in auto 1.39 in bios but shows 1.26 in windows.

super pi, 1m-14.765
super pi, 32m-15.55.297
Max temperature is 61C running dual prime95's.

So try some higher FSB settings, I could not even get into windows at 1400 but 1500 went right in.
This is all stable with 6 hrs prime95, and just finished playing 3 hours Call of Duty on line. I have the EVGA 122-ck-nf68 a1 board.[/img]

 

[EternalAlpha]

Speedfan: 39-45C (coolest core and hottest core); system temp is 33C
TAT: CPU0 temp is 60 and CPU1 temp is 57.

So the speedfan is giving me tcase = temp at heat spreader and values are 39-45C and anything under 50 is "good."

TAT is giving me tjunction = temp within the cores and values are 60 and 57 and anything under 65 is "good."

What's your ambient temperature?

I am interested in getting the Q6600, but I have heard that they can get pretty hot. That's a little worrying for me, as I live in a place where it gets very hot in summer.

 

[graysky]

What's your ambient temperature?

I am interested in getting the Q6600, but I have heard that they can get pretty hot. That's a little worrying for me, as I live in a place where it gets very hot in summer.

I can't remember what room temp was when I posted this. Lately here it has been 75-78F in the room which is I think 23C-24C or so? When I first built the machine it was closer to 65F in the room. If your room temps are that high, you'll definitely want to invest in a well ventilated case like the P182.

 

[graysky]

I edited the first post in this series, but decided to write it here too just to update the community:

Since I wrote this post about 1-1/2 weeks ago, I have lapped my ultra-120 x as well as lapped my Q6600. The decrease in temps have been amazing for about 5 hours of careful lapping, and $20 worth of sandpaper. The most recent temp data can be found the q6600 lapping post. I will say that after a few days of load/idle, the temps have dropped even more as the AS5 breaks in. Also, I discovered I don't need to run with a 1.310 vcore to keep the 9x333 stable. It's been stable now for >5 h on 2x orthos with a vcore of 1.3250 in the BIOS which is, according to CPU-Z and speedfan, 1.232 under load, and 1.260 @ idle.

I'll run the same x264 encode I did in the threads above with the new vcore tomorrow and post the results in a new thread.

 

[blkgto]

So whats the maximum temperature that you can take a q6600 to? Mine is at 38C right now in a Thermaltake Armor with a mini typhoon heatsink, only OC'ed right now to about 2.63. I would like to take it to 2.8 or 3.0.

 

[graysky]

Dunno about the max temp. I've had it up to 9x400 and it rebooted after about 50 minutes. I was ignorant about vcore when I did this and left it to auto. I also wasn't logging the vcore. I'm happy running 9x333 and having load temps under 45C -- that's really fast enough for me

 

[blkgto]

have you done anything with the vcore at this point or do you still have it at auto? Also what method of cooling are you using? I might give your way a try, just want to make sure my mini typhoon will do its job right.

 

[graysky]

Manually set my vcore 1.3250 in bios and temps have dropped by an average of 3.5 C (7 %) across all 4 cores. It's stable as judged by 2x orthos for > 5h doing small FFTs. If I drop it to 1.3000 in the BIOS, one of the orthos gives a math error after 40 min or so. I can obviously go back and select the intermediate voltage (1.3125) but this is fine for me.

 

[supaMASS]

I've been playing around with my Q6600 for a few weeks here are a few tips:

I don't know about the 680i borads but the P965's seem to have some FSB "holes." Here are a few numbers that I have found to work well.
330Mhz x 9
375MHz x 9
401Mhz x 8 (400 wouldn't post) 401 like magic
425MHz x 8 worked but haven't tested stability

All of the above with the exception of 425 (just haven't had time yet) have been stable running F@H for several days. Much tougher than prime95 and Orthos.

I'll post voltage settings, observed temps and more details later... (took forever to get registration email and I need to run)