Very nice indeed. :cool:
Printable View
Very nice indeed. :cool:
Excellent job with the cooling :)
You are getting closer to the same speeds as my ghetto waterchiller setup produces.
Must have been a fair amount of work involved.
I comend you on being able to do youre own modifications.
Instead of having someone do it for you:)
No doubt there must be a point of diminishing returns, where further cooling yields little or no improvement, but nobody seems to know where that point is. If we knew this, we could design for it.Quote:
i think you dont see a big dif in ocs with a -100°C cpu and a -200°C cpu... aynthing lower than -100°C would be a waste...
Here is something some people have learned from ln2 experiments with the latest P4 chips:
'dont keep the base (in other words the part of the cooler which is making the contact with CPU) of the cooler any colder than -120C'
Thanks. That gives us a target to shoot for: -120C/-184F.Quote:
Here is something some people have learned from ln2 experiments with the latest P4 chips:
'dont keep the base (in other words the part of the cooler which is making the contact with CPU) of the cooler any colder than -120C'
This would take a three stage cascade with something like R14 in the low stage, and something like ethane or ethylene in the middle stage.
Right.Quote:
Originally posted by LORD
But if it hit's -53.9 on the evaporator as set by chip-con then it will shut down the system. Or did you overcome this phenomenon?
Would be good if it can be circumvented with some sort of recalibration so as not to ERROR and beep.
I use 2 PSU one for the PC and one for the Prommy
First I start the prommy (and the Vapochill) when the temp going sub -53 the the errorbeb sound. Then push the left top button
the sound stop. Now start the PC.
If you use only R404 Prommy without cascade (or not -Bowman)
the temp may go higher than -53 and the display will work again. later the beep might sound again. Then just turn it of again. The PC remain on, and cooled.
You can calibrate -change the value on P14 to something lower. The display might show something -65 or -74 or something. It won't help the real temp under -53 will still activate the errorbeep.
PS: I have never before loaded the vapochill so hevy, that it won't go under zero. I then found out, that a modded PSU (380 Wat 12.3 volt) was to small.! The volt goes under 11.4 volt, and the Vapochill turned of.... I now use a lab powersupp. 12.8 volt together with the fan's and 2 Prommy (one for GRU) it takes almost 15 amp.
:slobber:
These "returns" aren't quite diminished.Quote:
Originally posted by Gary Lloyd
No doubt there must be a point of diminishing returns, where further cooling yields little or no improvement, but nobody seems to know where that point is. If we knew this, we could design for it.
:sick:
edit: OC by TAM, of course.
why did they say dont go under -120°C? what happened? no improvement in ocs or did the core crush or smthn?
I don't know what any of that means. I am totally clueless when it comes to overclocking, and you would not believe the dinosaur I call my computer.Quote:
These "returns" aren't quite diminished.
On the other hand, I have been trouble shooting and fine tuning refrigeration systems (from cascades to 3600 HP chillers and everything in between) for over 35 years, and have written several books on the subject.
Gary,
With those refrigeration skills of yours...you NEED a fast computer! Point is, performance INDEED increases with lowering of temp. This is a slightly outdated LN2 benchmark that the owner can probably outdo handily today. It seems that the best phase-change cooling in the hobby can only get a P4 stable up to about 4400mhz. With LN2 these same chips are going 10% faster at stable settings.
WOW! You should apply those skills of yours into a side business! VapoChill and Prometeia do a nice bit of business.
C
MAcci,
That's scary. That looks 99% like MY kid! ;)
C
Macci, what happens at -120c? I was planning on going much colder.
My side business is the books, which BTW would be extremely helpful to everyone in here, including those who are in the refrigeration industry. :D
I will soon be retiring and the books will be my only business. I have considered building low stage systems for people who have chillers they can use for the high stage. Who knows. I may do that yet if there is a market for them.
Gary,
Take a look at
http://www.chip-con.com/
it's pretty much the best out there...and they get $1000 for an upgrade model (like -55C at the evap)
Quote:
Originally posted by LORD
But if it hit's -53.9 on the evaporator as set by chip-con then it will shut down the system. Or did you overcome this phenomenon?
Would be good if it can be circumvented with some sort of recalibration so as not to ERROR and beep.
Just put a resistor in parallel to the temp sensor (use terminals 7 & 8 of the controller). This will increase the temp sensor reading.
The resistor value seems to be different for each prom. I settled for an 820kohm which gave a +5 deg C jump in the readings. This was enough to cope with the lowest load at idle (got down to -57 deg C ... readout was -52).
When you are manufacturing and marketing something, volume is everything. The rule is, go big or don't go. One of a kind mods are extremely expensive in terms of time. Bowman can tell you that he couldn't possibly charge for the hours spent. It is for him a labor of love, although he may eventually have something going with direct die block assemblies if mass produced. :D
CPU stops operating properly at around that temp.Quote:
Originally posted by FUGGER
Macci, what happens at -120c? I was planning on going much colder.
The older NWs (this is with 1.8A) could take a lot colder temps thou :D
http://www.solidhardware.com/macci/nwln2/7s.jpg
This particular 1.8A hit around 2850MHz aircooled, a bit over 3100MHz using -40C antifreeze and 3563MHz using LN2.
Nice setup.
Keep it well dusted,
At least I think you have no bugs, they will stick to the cold surfaces;)
If we you where closer to Denmark, I think we could take my 4353mhz an put your 5900ultra on it, that might give some nice 3Dmark scores:toast:
YOur system speed coupled with my video speed would indeed do some damage in 3DMark! :toast:
I think I need to get a new CPU thats not so mobo/FSB limited :D
ROFL.Quote:
At least I think you have no bugs, they will stick to the cold surfaces
That i have to see :D
tom*
du er seriøst for vild..
Thanks Tom Holck.
Oh yeah excellent results and cascade (Subcooling) mod :)
CodeRed, thanks as well. If it works as you've pointed out, then it seems a good option. Trick the sensor. :)
I think the max oc depends of voltage AND cooling. My new cascade R404 Prommy can go 4050 Mhz on default 1.55 VoltQuote:
Originally posted by ALT-F13
Tom, can you try to run your CPU on default voltage (1.55V i suppose) and check max clocks? Maybe not with this insane cooling though.
I'm asking 'cos my ES 3.2 works fine on 3.75G on default voltage on air (3.8G stable with waterchiller, +17C on chip) and no problems POSTing on 4.1G with that voltage... If i add any more it just get hotter and o/c worse (POST on 4.0, works on 3.7G even if it _actually_ on the same +17C). I'm waiting for Vapo now, but just trying to understand, WHY increasing voltage doesn't help me.
BTW when we'll see your 3dmark results with this monster?
I think 3.75 on air is nice, but 3.8 with waterchiller is bad.
Some ES chips oc realy bad. That might be your problem. Have You tried another chip (fx 2.4/800) to compair the system?
Looking forward to see what You can do on Vapo-cooling.
:toast:
tom
hit some buttons to get the beep away ;)
i have a m1 doing -57 on 507
P25=-50 (low temps alarm) mine = -60
http://www.tweakers.net/ext/f/9568/full.jpg
this display is more adjustable then the m2
i have a slot version vapo en a classic also
maybe i can use my m2 to cool te m1 :)
that doesnt stop the alarms on mine.Quote:
Originally posted by petervandamned
tom
hit some buttons to get the beep away ;)
i have a m1 doing -57 on 507
P25=-50 (low temps alarm) mine = -60
P25 is set to -99 for me
the only thing that worked was the resistor mod, just a bit hard to calibrate.