Kentsfield/Quad-Core Overclocking Thread

[Jon]

Take a look at the Thermalright Ultra 120 Extreme. Best aircooler out there!

From what I've read, its too big to fit in the thermaltake armor cause of the 25cm side fan...

It has got excellent reviews though.

---------

Also, from what I've read so far, with the Q6600 the max of 62 degrees is for the heatspreader. I don't want to find out, but from what I've read, the cores don't begin to throttle until 100 degrees. Anyone for fried eggs?

---------

Just ordered the Zalman 9700... now just have to wait until Tuesday...

Actually slaWter, I was hoping you could help me with a 9700 installation question (I notice you have one already in your setup). The flash movie on zalmans website shows them fitting it using the v shape bracket. They hook on one side then use a screwdriver pushing down to hook over the other side. But in the online manual it says to screw down each side a bit at a time with the provided screws. Which method should I use?

[walterman]

So, the heat is a huge problem for quads, right ? Must be very hard to keep those overclocked cpus running cool.

[Jon]

Nice, looking at the photo, I'm glad I went with the Zalman (only just clears my side fan by about 2cm).

---------
Edited:

Ok, thanks for the advice guys, I understand what you were telling me now... especially the 1:1 memory thing. It will not work at all past 2.7ghz if I change the memory multiplier to anything but 1:1. I was having freezing at 2.7ghz when it was previously stable because the bios reset after a crash and set my memory timings back to default (3x) making my ram run at 900mhz instead of 2.5x (750mhz).

In conclusion... I set ram multiplier to 2x, all voltages stock... upped the fsb 10ish at a time until I hit 333. I'm currently running at 3ghz (memory at 666). OCCT says its stable... Fear benchmark was 100% >40fps... and after an hour of Rainbow 6 I am really really pleased. I can't believe these quad cores clock so well!

Oh, and now I get why having faster memory won't really help me... I assume the 3x multiplier will make the system as unstable as the 2.5 multiplier.

Oh, BTW... temps are not too bad compared to some I have seen. Max I got all night was 64/65 degrees with all 4 cores 100% loaded for 2 hours. Using Core Temp to measure. And the room was quite hot cause I didn't have the air con on at the time! In comparison it idles at 40 degrees with air con on, 50 with air con off.

[walterman]

How many watts do you need for a quad core ?

I was testing that PSU calculator at (http://www.extreme.outervision.com/index.jsp), & with a 8800GTX it told me that i needed >1000W

Dunno if it's real, but seems overkill for me.

[Brent101]

Originally Posted by walterman

How many watts do you need for a quad core ?

I was testing that PSU calculator at (http://www.extreme.outervision.com/index.jsp), & with a 8800GTX it told me that i needed >1000W

Dunno if it's real, but seems overkill for me.

Under system type, make you sure you have "single processor" selected, even if it is a quad core cpu. If you have it set to quad processor, you are telling it that you have 4 phyiscal cpus/sockets. Im pretty sure that is why it is saying 1000+ watts.

[einstein_314]

Ok guys, here's a bit of an update. Currently running my Q6600 at 3.6 (400x9) @ 1.5V (in bios...windows reads 1.4V). Temps seem to maxing out at 58C. Seems to be stable so far. I have a question though....How do I change the stupid multiplier on this P5B-Deluxe? I change "Modify Ratio Support" to enabled and then change the multiplier...but it doesn't actually change. Am I missing something?

EDIT: Ok so turns out it wasn't actually at 3.6.....because the multiplier was being all wonky. When I change the multi in the bios, the clockspeed is not being correctly reported in windows...or on the POST screen. They show that it is the FSBx9 (the default multi). Everest seems to pick up the correct multi though. Weird stuff. So for now I'm back down to 3.2. I'll have to increase the voltage some more to hit 3.4 or even 3.6. And I was so excited to get past 3.2. Oh well.

[DRen72]

I did not think you could change the multiplier on a Q6600. I thought it was locked.

[schuey74]

Originally Posted by DRen72

I did not think you could change the multiplier on a Q6600. I thought it was locked.

You can change down, not up. At least that's the way it's worked on my P5B-DX with several different CPUs.

[mhw100]

I currently have a Zalman 9700. It's not bad but as others have pointed out it isn't that great compared to Thermalright's IFX or Ultra Extreme. The problem is the tests I've seen have been only on C2D not on Quads.

What I've seen reference to is that the heavy Thermalright products weigh down on some of the cores thus giving a heat disparity between them...sometimes as great as 15C. For example by simply lifting the weight of the Ultra off of the board with you hand will equalize the temp.

Therefore, even though the cooling of these Ultras are almost as good as water on C2D, I wonder if water is the only solution so that the heat dispensation is equal on all cores?

Comments?

[einstein_314]

Originally Posted by mhw100

I currently have a Zalman 9700. It's not bad but as others have pointed out it isn't that great compared to Thermalright's IFX or Ultra Extreme. The problem is the tests I've seen have been only on C2D not on Quads.

What I've seen reference to is that the heavy Thermalright products weigh down on some of the cores thus giving a heat disparity between them...sometimes as great as 15C. For example by simply lifting the weight of the Ultra off of the board with you hand will equalize the temp.

Therefore, even though the cooling of these Ultras are almost as good as water on C2D, I wonder if water is the only solution so that the heat dispensation is equal on all cores?

Comments?

First off, welcome to nvnews!!

I'd never thought of the effects of the weight of the heatsink. But I suppose that would make sense. The weight is trying to pry the heatsink off the chip and thus the top has less contact than the bottom. Whether it is significant enough to notice (you say 15C....is this just a guess or a number you got from testing?) I dont' know.

But I will say that water is awesome on my chip. Maxes out at 59C. And idles around 35-42 depending on whether I turn down the fans on the radiator or not. Normally I run the fans at low (can't hear them at that setting) and the temps are fine for normal use. As soon as I start doing something CPU/GPU intense (like a game) I crank the fans.

[mhw100]

Originally Posted by einstein_314

First off, welcome to nvnews!!

I'd never thought of the effects of the weight of the heatsink. But I suppose that would make sense. The weight is trying to pry the heatsink off the chip and thus the top has less contact than the bottom. Whether it is significant enough to notice (you say 15C....is this just a guess or a number you got from testing?) I dont' know.

But I will say that water is awesome on my chip. Maxes out at 59C. And idles around 35-42 depending on whether I turn down the fans on the radiator or not. Normally I run the fans at low (can't hear them at that setting) and the temps are fine for normal use. As soon as I start doing something CPU/GPU intense (like a game) I crank the fans.

Thanks for the welcome.

The 15C is from a 2 independent sources which doesn't make it scientific or statistically significant but makes one ponder the matter?

I was kind of hoping that water would cool me way down but your 59C on water seems high to me with you running at 3.2Ghz. I'm at 3.44Ghz -at idle its ~31C at orthos load ~55. 65C is the ceiling so we're not far from it.

Also, I'm trying to locate some empirical evidence comparing a good air setup to water. Do you know of any such studies? I see a lot of anecdotal fluff or cheapo air setups compared to water but nothing substantive.

[schuey74]

Originally Posted by mhw100

Thanks for the welcome.

The 15C is from a 2 independent sources which doesn't make it scientific or statistically significant but makes one ponder the matter?

I was kind of hoping that water would cool me way down but your 59C on water seems high to me with you running at 3.2Ghz. I'm at 3.44Ghz -at idle its ~31C at orthos load ~55. 65C is the ceiling so we're not far from it.

Also, I'm trying to locate some empirical evidence comparing a good air setup to water. Do you know of any such studies? I see a lot of anecdotal fluff or cheapo air setups compared to water but nothing substantive.

An overclocked quad core and a GTX in the same loop is going to stress most watercooling setups, especially when you start bumping up the volts. If his GTX was off the loop his load temps would drop around five to seven degrees(c).