Comp

[cnumartyr]

Hey man, P5K-E finally came in the mail today.

You were right, I LOVE it. Voltage Damper is great. 8mV vDroop under load at 3.6 GHz!

 

[CompuTronix]

Isn't it just great? 8mV is typical on mine, and the max I've ever seen at ANY setting at ANY load is 16mV!

Comp

 

[cnumartyr]

It is most certainly a nice change. Didn't even reinstall Windows.

I had one brain fart though and couldn't figure out why it wouldn't POST at 3.6 GHz. Then I remembered that this isn't quite like the DFI and that there is some droop between the BIOS setting and actual... Once I remembered that it easily ran up to 3.6 GHz without much stress.

 

[cnumartyr]

I love the board, finally hit 3.6 GHz stable!

 

[CompuTronix]

Nice work! I think your G0 is slightly cleaner than mine, since I need 1.44 Vcore for 3.6 Prime stable, so my temps are 60c CPU and 72c hottest Core at 22c ambient. Also, your running a better cooler, and you've lapped your CPU. Your Idle and Load temps are excellent as well as accurate! What is your ambient?

Comp

 

[cnumartyr]

Nice work! I think your G0 is slightly cleaner than mine, since I need 1.44 Vcore for 3.6 Prime stable, so my temps are 60c CPU and 72c hottest Core at 22c ambient. Also, your running a better cooler, and you've lapped your CPU. Your Idle and Load numbers are excellent!

Comp

I personally think I can do better... I was sooo tired when I was doing putting it together last night that I just did a halfway job on the CPU/Heatsink.. I'm going to go back over them as I only did them to 800 grit, got a bit angry and didn't get them completely flat... I'm not satisfied with how I did. I also rushed getting the heatsink on, so I think I could do better there.

All in all I'm pretty happy with it though.

Edit: That wasn't a true idle, it was about 45 seconds after stopping Prime95.. however it's about what they are now as well. Ambients were 22 C when I left this morning and 24 C when I got home this afternoon. Room was closed up and the heat was on to the house (Thermostat is in the hallway outside the room, but door was closed).

 

[cnumartyr]

So it came with 0703 BIOS.

Just flashed it up to 0602 and wow...

vDrop (BIOS Setting vs. Actual) went from 111 mV to 24 mV. Feels like I could squeeze a lot more out of this thing now... I was worried I wouldn't be able to go over 1.6v!!

 

[CompuTronix]

I think you'll find that once 1.35 Vcore is reached during overclocking, each 100Mhz increase in CPU clock speed typically requires a corresponding increase of 0.05 Vcore, which results in a 3c to 4c temperature increase. This behavior is very linear and consistent, so based on your temperatures, you can easily run 3.7Ghz. If you wanted to push your Vcore to 1.5, then I'm sure you could run 3.8Ghz.

Personally, I think the sweetspot is 1.45 Vcore, so by all means, crank her up a little higher. Perhaps 3.75Ghz at 1.475 Vcore with a ~6c increase to ~67c Core temperatures might represent a reasonable compromise for your rig.

Comp

 

[cnumartyr]

Sounds pretty good.

Working on my memory and FSB right now. Don't want to OC the CPU past 400 MHz FSB without knowing for a fact that the NB/Mem are stable (makes things a pain to track down).

Gotten the memory up to DDR2-1000, I'd like to hit 1200 (ambitious.. haha). With the FSB I'd like to see 450 or so atleast. We will see where I end up.

 

[CompuTronix]

With referrences in auto, and term, NB and SB at minimums in manual, you should be stable at 438, however, at 450, the FSB needs 1.4. I've pushed the FSB to 475 x 8 for 3.8, but it requires so many BIOS alphabet-soup tweaks, that it's not worth the trouble when I can run 422 x 9.

 

[cnumartyr]

Running 32M SuperPi on DDR2-1200 2.1v 5-5-5-15 right now.

If it passes I'll memtest it tonight from a Boot CD after I blend/play games on it the rest of the night.

Edit: Who needs Ballistix... lol.

 

[CompuTronix]

That's amazing memory overclock for PC6400. Hope it works out for you.

 

[cnumartyr]

That's amazing memory overclock for PC6400. Hope it works out for you.

Had to move to 2.2v to get it 32M Stable. Running Blends now.

 

[cnumartyr]

2.3v+ Will get it stable at DDR2-1200. I just didn't want to run that overnight just yet. Here's the blend overnight on DDR2-960.

That will get me up to 480 FSB 1:1 so it's good enough for now. 11 Hours of Blend and Speedfan Temp logs.. and here is the result!

Max Core Temp: 65
Min Core Temp: 31
Average Core Temp: 55.92
Max Tcase: 53
Average Tcas: 46.5

Cores 0/1 ended up with the same average: 57.38
Cores 2/3 ended up with almost the same: 54.48 / 54.47

 

[CompuTronix]

Excellent! What was your Ambient?

 

[cnumartyr]

lol, not sure how I forgot that one, but:

Ambient was maintained throughout the night at 22 C +/- 1.

 

[CompuTronix]

You're just a few degrees warmer than I estimated, however, you had said that you were going to revisit the lapping task. Regardless, your Tcase to Tjunction Delta is 12c under Load, which is exactly the same as mine, so if you pushed up to 3.75 at 1.475, then your temperatures would be about CPU 59 / Core 71, which at 22c ambient doesn't exceed the top of Safe Scale. Mine runs 3.6 at 1.44 and CPU 60 / Core 72 with my lapped ACF7Pro and X23 under Small FFT's, so you're doing great!

 

[cnumartyr]

So I can't 100% compare the results.

However here is a 30 minute log from SpeedFan for a Small FFT test. Ambient was actually up to 24C. This is after reassembling and doing the lapping properly.

Max Core Temp: 60
Min Core Temp: 48
Average Core Temp: 56.85
Max Tcase: 49
Average Tcase: 47.36

Pretty happy with the results having redone this.

Idle Tcase has also dropped 1-2C on average.

 

[CompuTronix]

Cleared for flight level 375

 

[cnumartyr]

Cleared for flight level 375

And that's with the old fan, wanted all variables I had under my control to be the same.

Here are the new fans:

And for fun, here's the heat sinks I added to my old Blood Iron to give to a friend:

 

[CompuTronix]

Very nice work. I think you read my configuration info... the entire rig is cooled by a single 120mm 100CFM intake at 2500RPM, and it does quite well, but I wonder if a 130CFM would drop the CPU a few more degrees?

 

[cnumartyr]

Very nice work. I think you read my configuration info... the entire rig is cooled by a single 120mm 100CFM intake at 2500RPM, and it does quite well, but I wonder if a 130CFM would drop the CPU a few more degrees?

I'm not sure really.. Part of the problem I'm seeing is air diverting out of the TRUE rather than into it and through it. Part of this might be static pressure issues. The reason I plan to use the push/pull is to cool the "back" heat pipes more effectively.

I don't have a side exhaust, but I have holes for one over the CPU. I can feel air exhausting out the side from the fan trying to push air through the TRUE.

I really picked up the Ultra Kaze's to do some CFM testing and Static Pressure/Noise testing. I'd like to use them on a Radiator if they aren't TOO loud if they really do push a lot of air.. Only thing now is it has a big dead-space in the center hub.

Your configuration is great. However I too would wonder since you are creating slightly positive pressure if a higher speed and static pressure fan would increase airflow and perhaps drop temperatures slightly. I think it's an interesting enough idea that I would pick one up to try it out.

 

[CompuTronix]

I plan on doing some minor upgrades over the holidays, which includes changing out my ACF7Pro (126.5mm) for a ZeroTherm BTF90 (128mm) - http://www.newegg.com/Product/Product.aspx?Item=N82E16835887002. I think it will just squeeze into my case, which is limited to 130mm, and the thermal benchmarks are better in the reviews, so with a fresh lap, I think I can shave off another 3c, and the UltaKaze might compliment the new cooler well.

 

[cnumartyr]

I plan on doing some minor upgrades over the holidays, which includes changing out my ACF7Pro (126.5mm) for a ZeroTherm BTF90 (128mm) - http://www.newegg.com/Product/Product.aspx?Item=N82E16835887002. I think it will just squeeze into my case, which is limited to 130mm, and the thermal benchmarks are better in the reviews, so with a fresh lap, I think I can shave off another 3c, and the UltaKaze might compliment the new cooler well.

With all those heat pipes it's almost like the limiting factor will be the airflow made by the fan!

The Kaze might certainly help get the air there a bit more efficiently. We both know that as the pressure difference across the fan increases it's airflow will decrease... The more positive static pressure inside your case the less your front fan will flow. I think the Kaze will flow more air at a higher static pressure than what you have.

On that note, what do you have up there right now? What are the odds it's a Cooljag Everflow?

You'll have to post up some pics of the rig, I'd love to see it.

Edit: Btw what are your thoughts on watercooling the beast you got there?

 

[CompuTronix]

Odly enough, I've never taken any images of my rig when I've had the cover off, but I will certainly make a point of doing that over the holiday upgrade. As for liquid cooling, I've considered it many times, but there just isn't any other space on the boat to install or position the radiator in the compartment. Also, I occasionally pull the rig out of the compartment and take it to friends houses for Flight Simulator sessions, so mobility is a factor.

Here's a FrostyTech review which covers a wide variety of coolers - http://www.frostytech.com/articleview.cfm?articleid=2045&page=4 - which shows the BTF90. The performance appears to be a few degrees better than the ACF7Pro, which is not shown, but the ACF64Pro is the same cooler with a different mounting bracket, so I think the comparison is valid. Also, unlike the ACF7Pro, the BTF90 has a backing plate.