what block on naked PhII X4 940BE ?

[Tommie_lj]

im using a pretty old Aquacomputer Cuplex Pro cpu block now, but i dont like the temps im getting with it..

my loop:

Phenom II 940BE @ 3.7ghz 1.5V
Aquacomputer Airplex Pro 360 rad
Aquacomputer Cuplex Pro cpu block
Ultra Kaze 2000 at 5V
Laing DDC 10W
homemade res
3/8 barbs and tubing

temps:

CPU: Idle, 32-35C (C&Q activated) Load, 48C
Water: Idle 25C, Load 29C
Ambient: 23C

(approxiamtely) since i have no idea how accurate these sensors are... but at coldstartup the watertemp is exactly the same as ambient.



picture of the chip

[shazza]

Not sure you'll see a lot better performance with a different block - it's hard to find info on water blocks and AMD performance.

What Radiator and fan combo are you using?

For comparison purposes, I'm seeing about 40 degrees on the PII 955 with the Alphacool Nexxus Highflow block (which I don't think is anything special) and a Magicool Extreme 240 rad with Low Speed Yate Loons - ambient 23-24 degrees C. (note - this is in an open setup on the Torture Rack, so probably a few degrees better than I would see enclosed in a case).

[Tommie_lj]

updated the info a bit.. im running kaze 2000 fans at 5V, radiator mounted outside and on top of the case.

this radiator looks exactly the same as swiftech MCR 320..

[Zeus]

Little offtopic but how exactly did you remove the heatspreader?

[Boogerlad]

probably sanded it off.

[Tommie_lj]

yupp, sanded 80grit til youre nearly throug, 240 til youre through, then finished it with 600 and 1200 grit to get it smooth.. you have to cool down the chip after some sanding, used the bottom of a "cooking pot" (or whatever its called in english ^^ ).. just put the cpu upside down on it for a few secs, and its cool again..

took 7 hours ++ , but it was worth it gave me a 1c drop in temp.. lol! hoped for a better result, but after all.. its alive and it didnt make it worse

[Jah]

How about a Fusion with a small nozzle, should give you concentrated cooling on the die, any old school single core block should be better optimized for a small cooling area as far as I understand.

Hi BTW, lideri

[Razrback16]

Upgrading rad/fans will help more than a different block probably. I'm currently running a Swiftech Apogee GT. I don't know what my idle/load temps are though, I don't run an active monitoring app on it in Windows. BIOS says it idles around ~30*C.

[ecat]

I love its nakedness

Comparing it to my clothed 940BE @ 3.8GHz, 1.43V, allowing for 19C ambient, my 2x pa120.3 set up, more powerful pump, temperature controlled fans, no C&C and recalling the numbers from memory ( ) I'd say your temps look about right ( ). You may gain an extra 1 or 2C at load with something like the HK 3 I use.

One thing that would worry me would be the convex nature of the base on some blocks, I'm unsure how the naked cpu die would cope with the localised pressure as the block was tightened down.

Other than that, you could add more/faster fans and/or another rad. With 2 x pa120.3 + 6 YL @ 800 rpm my water load-idle is seldom more than 1.5C, this should give you another 3C, maybe slightly more, at the cpu.

For reference:
Amb: 19C, cpu idle 25C, cpu load 36C, water idle 21C, water load 23.5C

[Tommie_lj]

Ecat: even if i bring the water temps down to lets say 25C at load, it will only give me 4C cooler cpu, at most.. right?

Jah: tjena, the fusion block looks very interesting..

to all: i dont think the rad/fan setup is my problem here, and not the pumping power either.. running my Ultra Kaze at 2000RPM only gives me 1C improvement from running 5V... and modding the pump to 18W didnt make a difference at all..

but i may test modding the block a little.. there is som sot of "accelerator plate".. just a tiny plate with small holes in it.. i want to replace this with a piece of tubing with a narrow hole.. making extremely pressured water hit hte center of the block.

[ecat]

Ecat: even if i bring the water temps down to lets say 25C at load, it will only give me 4C cooler cpu, at most.. right?

With the block you have, yes, 4C is a big number in this place . With a better performing block I'd expect an extra couple of C at cpu load. It's all to do with the sum of the parts.

[Tommie_lj]

Ecat, 4c is a big number, but the price to achieve that is also big

http://www.hardwarelabs.de/artikel.php?id=2208

internals of my block. do you think there is room for imporvement on the accelerator plate? using something like the fuzion maybe?

[Tommie_lj]

im going to try a new "injector".. what size do you guys think may be suitable on the hole? (hope you can understand my swenglish )

[_G_]

incredibly still in stock, the legendary swiftech storm.
http://www.jab-tech.com/Swiftech-Sto...2-pr-3022.html

[Tommie_lj]

where can i find tests on this block? do you think modding the pump to 18W with that block will help?

[_G_]

don't think there's any current data compared to current blocks but its main problem is it didn't perform as well on IHS cpu's (it was tailored to small contact areas)


edit: its a restrictive block so stronger pump would be a plus
nice work on lapping the ihs off that thing, waaay more patience than i have lol

[Tommie_lj]

but will it perform better than my current block? that is the question and how much..

thank you.. it was eorth the effort even if it didnt help temps much.. naked chips is sooooo sexy