Cooling Persivore and Highland3r's bench session

The day was a partial success - really wanted a 100% overclock from the Celly, we'd have gotten it but it seems the board wouldnt boot at 266 FSB or above, we're assuming due to a change in chipset strap here and without a version of clockgen we were unable to clock from windows.
The day was about testing and having a bit of a play more than anything. It's the first time either of us have dealt with dry ice and we've learnt LOADS.
It's MUCH more fun than benching with a single stage, there's another angle as you have to watch the temps all the time adding more pellets as required.
It's also a shame we didnt get chance to freeze some GPU's but the tube didnt make it back in time

Utterly shocked at how well the P5B clocked too. Persivore didnt do a chipset or MCH mod to the board so everything was running stock on that front. Still managed to pull 500 FSB with ease, would probably go further too had we had the time to test.

So, onto the results -

This was the best pi 1m time of the day. So SO close to 10 secs, we just ran out of time at the end and couldnt get a run under 11 Have just noticed that this was run at the same speed as the 32m run so theres certainly room to improve there!


32m run. Just what it says on the tin!

Due to time limits we didnt really get chance to do much more than this today. As I said above, it was mostly about learning a new cooling method. There's a little ICE left, so might have another stab at things tomorrow.

Temps wise, we think the contact when running the Conroe was a little poor. We saw -56 in bios initially, and this was down to ~ -47 after a few hours. Temp probe stuck to the base of the tube read ~ -67-68 under load and -70 idle.

As for ozstrikes request, there's no ice pics. Have some of a smoking cup of coke though, will dig it out for you. Condensation was non existant also. Even after a good 4 hour straight run (removing the tube twice during this time) there wasnt an ouch of condensation. Everything was dry as a bone, just what we like to see

Sorry for the lack decent pictures and interesting results. 4.6ghz from a conroe, 500 FSB from a sub- 100 GBP mobo and 5+ ghz from a celeron aren't bad for DICE nubbins....

 

Nice, is there anything you ccould do to get the celly up to 100%
Hmm would like to mess round with that kind of thing.

 

Clockgen would be good. Possibly another mobo

 

Some pics from today... (sesh #2)







Temps are MUCH MUCH better than the last session. We're seeing ~ -46 in bios @ 4.68ghz/1.78 vCore now, compared to low 20's last time around.
Also to add, ran some benchies yesterday with isoproply alchohol and temps were to be frank shocking. -60 at tubes base, with -22 in bios @ ~ 4.65ghz. So, in short isopropyl sucks for dice

Speed wise, we've hit the max of my E6600 on sensible voltage, just switching over to Persivores now so fingers crossed we'll go a bit further.

 

Mixed bag today. Temps were awsome think we've finally got contact etc licked with the Chilly.
Both CPU's maxed at pretty much the same place in the end, could well be mobo limits perhaps. That needs some more investigation.
The "other" motherboard mentioned was Abit AB9, also 965 chipset but doesnt run 1333 strap over 400 FSB. At 450 it was 1/3rd sec quicker than the P5B with slower memory. However it can't match the P5B for overall FSB.
Needs some work on chipset cooling it seems.

Still have ~ 20kg (2 bags) of Ice left should be enough for some more runs later/tomorrow all being well.
As a note, 30KG is FAR too much for 1 days worth of benching. Based on last time though when we lost ~ 1/3rd before the stuff even arrived I thought it'd be better to get some extra.
Cold weather seems to have helped us out somewhat this time.

http://www.yara.co.uk is where the Ice came from, they'll ship UK wide for about 25 quid.
It's 14.20 for 10KG (+vat) and 6.50 (again + VAT) for a fish box. Fish box will take 2 10kg bags.
If you only want 10kg, then its worth paying extra for a better box as sat on its own in the normal fish boxes you'll loose quite a bit of ice.
Also, as mentioned earlier don't use isoproply alchohol. Temps are ~ 10-15 deg worse than Acetone.

Results:

and


We had a better run with Persivores CPU, but he can have the pleasure of posting that one!
Need to get some 3d done to shut Kenny up now (love ya really mate )

 

good to see you playing with some Dry Ice

 

The lowest temps I've seen are actually on a brazed tube with LN2. That's because you want the thinnest wall possible. They used a low profile, thin, xeon heatsink, cut the fins off, and brazed a super thin copper pipe on top of it. They hit -193.2*C on LN2.

Also you may want to try useing less ram rather than more. I've heard reports of ram raising temperatures when capacity is increased. I know this is true on opty's and I also think most overclockers (when going for max GHZ) stick to 2x256mb of Fast ram. But stick the 4gb's in for any 3dmarks and such

 

Is it realy true you havn't modded the board.
My exact same board hit a fsb wall at 370 mhz.
Now i feel bad cus the board can do a lot more than I managed.

But your result do look good. BTW. Can you post some performans results from the Celly at 5 ghz.

 

Thin walled tubes arent ideal for load holding. Gives quite a bit of temperature flucuation. Would rather have a tube which doesnt pull as lower temps and holds them rather than uber cold but without the capacity to maintain them.

We only ran 2gb at anyone time ram wise, sorry wasn't clear from the original post. Need to get a 1gb kit sometime and see how that affects the overclock though

The P5B has vMCH and vFSB mods done however we didnt need them until ~ 520 FSB.
Try booting at 401 FSB, the chipset should run a looser strap there. Your CPU should make it to 401 without too many issues, may require a small voltage bump perhaps.
Also make sure you've got the latest bios running as this helped the max FSB on this board (in fact we're still running 0509 bios iirc, which isnt quite the latest)

Sorry, we didnt screengrab the Celly @ 5ghz mate, will try and get it for you next time.


Contact even better on the final run yest, thats running ~ 1.47 vCore as well, highest temps went @ 4.7ghz and 1.8v was -48. Lovely

 

Thanks for the help Highland3r.
Made it to 470 mhz fsb.
But further it won't boot windows. Do you think a jump to 501mhz fsb could do it or was that just working from 370 mhz to 401mhz?
And now the dry ice would come handy. It's is hitting a worrying +75C Is this dangeorus?

 

Nope, from 401 upwards the board uses the same chipset strap. Could well be your CPU maxing based on those temps, or the memory. Maybe drop the multiplier and see if you can push any further with lower CPU speed.

 

Ok. Well thanks anyway.
Maby it's a new cooler needed. (Idle temps stock at 50C+)

Dropped the multiplier but couldn't get the fsb higher. But I'm happy were I am. 3,3 Ghz is enough for me!

 

Found a screeny on portable hdd mate, here you go

 

Thanks for that.
It looks to perform decent when having in mind it's a celeron!