NCE Build

[pythagoras]

Spoke with Scott earlier today, it sounds like everything is going great except for the temps Just need to figure a way to keep the temps from crashing down to -135 to -140C after a benchmark and locking up his QX9650.

Load F@H. It will idle when benching and then use 100% of all four cores when you stop benching. Should be an instantaneous switch.

Regards

John.

[n00b 0f l337]

Hows load holding when ya bench?

[n00b 0f l337]

Any pics of ice and such on mobo? Would be great to see.

[n00b 0f l337]

HEY! WCG.. and configurable for percentage of cpu load!

Hahhaa sorry Dave but It just doesn't pickup like F@H does at least for me. If you have four instances of F@H; one on each core; the pickup of heat is so quick at end of bench. It actually picks up at end of all benches and when its calculating score before score is shown.

[[XC] gomeler]

Well I can guarantee you running WCG or F@H for the science itself wouldn't work, they'd have so many corrupt results

Here is an image of the mobo frozen up.


I don't know why I didn't think of a crunching app, so dumb of me as that's what I always suggest to prevent floodback *doh*

[[XC] gomeler]

As for how it holds a load, you'll have to ask Scott but from what I gather it was doing fine at 5200+MHz. I'll be prepping one of my DQ6's for 3 stage benching this week, overkill insulation is in order.


Here is a better image. My camera is nice but it is still a point/shoot so focus is a pain sometimes.

[[XC] gomeler]

HMM, Charles ran it at 5400 on his Yorkie ok..

Heh, I'm just saying I don't think in the time not spent benching it could complete a result, submit, validate, and actually pass quorum Using it to keep the processor warm = good. Actually submitting potentially badly flawed data = bad.

[n00b 0f l337]

Oh they've got plenty of capacity I'm sure. That thing can hold a cold evap sub -100C for quite a while without even the compressor on

[[XC] gomeler]

Oh they've got plenty of capacity I'm sure. That thing can hold a cold evap sub -100C for quite a while without even the compressor on

Not with a load, we tested it

[[XC] gomeler]

good point, but a yorkie at 5400m might do a dengue fever WU in 4 mins..

It would be fun to run the BOINC benchmark at some insane clockspeeds, will have to keep that in mind.


Did you sleep for 3 hours or something? Could have sworn you went to bed I screwed up bad with a nap from 8pm-midnight

[Movieman]

It would be fun to run the BOINC benchmark at some insane clockspeeds, will have to keep that in mind.


Did you sleep for 3 hours or something? Could have sworn you went to bed I screwed up bad with a nap from 8pm-midnight

No, built a machine, took out the rubbish, got the old lady breakfast and I'm still here working coffee #2

[[XC] gomeler]

No, built a machine, took out the rubbish, got the old lady breakfast and I'm still here working coffee #2

Alright I'm done spamming this thread, time to grab a shower and some breakfast. Need to wake up, lots of parts to track down today for a mini version of the NCE

[sdumper]

General question: does anyone know of reasonably priced K type probe reader which can measure temps in a range of -200c and warmer? I cant find a replacement for my CEN-TECH 92242...

Back on topic...Im playing a bit of hooky from work today and working at home...hehe

Anyway in between emails I am messing with the NCE and added a thin layer of tin foil between the cpu and shims which has given me about 20 seconds longer before I hit my coldbug shutdown.

3DMark06 either wasnt stable at 5.3 ghz or I got too cold because it I froze on the first sequence and then shut down.

It takes me about 90 seconds under somewhat of a mild load to drop from -130 to -100 when I shut the 3rd off. How can I determine the pressures have stabilized and its safe to kick on the 3rd stage again? I might be able to run a heavy load bench and shut off the 3rd stage just before it completes so that I dont coldbug that way I can at least show a score of wprime.

In hind sight I really hate the fact that the mother board doesnt read temps and im flying in the dark

[sdumper]

Moderator's can we move this out of vendor relations?

All has been resolved and the unit is delivering well beyond my expectations.

[[XC] gomeler]

The UEI DT150 and DT200 are the cheapest K type readers that I know of. They aren't a fluke, that is for sure, but they'll get the job done.

Glad to hear the aluminum foil is working, maybe try two layers and progressively build up a stack? I'm hoping that you can detune the heat transfer between the evap and CPU enough to get the IHS to idle around -110C or so as it seems the chips were coldbugging when the evap was at -135 to -140 so I imagine the IHS was around -120C.

edit: with the pressure equalization if you could isolate the noise you could listen to the suction line. When you shut it down you should hear gurgling noise as the system equalizes through the capillary tubing.

[Yellowbeard]

I wonder if something like aluminum flashing or similar thin aluminum materials may work as well? For some reason, I don't like the idea of multiple layers. Once you reach a desired thickness, it seems that a solid piece of the appropriate material would be best. But, I certainly am not a "noob of leet" sub zero cooling guy, just the noob part.

[RyderOCZ]

Moderator's can we move this out of vendor relations?

All has been resolved and the unit is delivering well beyond my expectations.

Just my 2 cents..

Start a new thread in Vapor/Phase with questions about controlling temps

[[XC] gomeler]

I'd say another thread with the results would be great and then close this thread with the last post linking to the new thread. That way it'll get more exposure and I imagine the phase change guys would like to see the fruits of all your labor.

I wonder if something like aluminum flashing or similar thin aluminum materials may work as well? For some reason, I don't like the idea of multiple layers. Once you reach a desired thickness, it seems that a solid piece of the appropriate material would be best. But, I certainly am not a "noob of leet" sub zero cooling guy, just the noob part.

The problem isn't the thickness but rather the heat transfer. He's got 1/4" copper blocks and the problem is he's only putting 2 "jumps" for the heat transfer to cross. I was hoping with the aluminum foil to add more of these jumps where the heat transfer is less efficient. Super tiny gaps are terrible conductors, even if there is thermal paste in between each sheet. With his 1/4" shims he was still coldbugging. It's nuts to say the least.

[sdumper]

http://www.xtremesystems.org/forums/...84#post2900984