PDA

View Full Version : Our first Dry Ice sesion - E6700 @ 4740Mhz

Monstru
08-07-2007, 01:10 AM
A few weeks ago, me and my friend Matose were trying to get something out of some Connies on air, while having +46/48 degrees Celsius ambient temperature. I became obvious pretty fast that we needed something else, so we went to our gurus (M1cutzu, Sang and Xwoodoo) asking for some help. M1 suggested that we should try some Dry Ice action, to get used with sub-zero benching. Said and done.... M1 and Sang provided us with the dice pot (full Alu, nothing fancy) and with lots of useful advices, tips and tricks, things that would have taken us days to dig out from all the related threads here at XS. On the other hand, Xwoodoo provided us with reliable Ram (Mushkin Redline XP2-8000), with a strong PSU (Mushkin HP-580AP) and an air conditioned location. Sang gave us the insulation and lots of advice, that we really found useful during benching. So, having full support from "the big guys" here at CrazyPc, we started gathering everything that we needed to start the action.

The Platform:

Motherboard: Dfi Infinity P965 , SilenX NB cooler, no mods.
DDR: Mushkin Redline XP2-8000
PSU: Mushkin HP-580AP
Video: Ati 3D Rage PCI :D
HDD: 40Gb Sata 2Mb Buffer


CPU's

Intel Core 2 Duo E6700 (FPO L627B)
Intel Core 2 Duo E6600 (FPO L639A)
Celeron D346 (Prescott Core)

We used nail polish to insulate around the socket and on the back of the mobo. We also use armaflex for the insulation.

Obviously, first time we tried to start the mobo after we had mounted the pot...it wouldn't start. We lowered the pressure from the top plate, and we heard that sweet sweet sound that every overlocker loves to hear after pushing it to much: "beep"...the mobo works.

E6700 - This one was very hot, and our Alu tube was pretty thin, so we couldn't get it higher that 4740Mhz (Spi Stable), vcore 1,71v, @ -52C. Still, we consider this to be an ok result for starters, with a CPU that doesn't go higher than 5,1 on LN2... Result: 10.687s

E6600 - The minimum temperature that we got was -62C on this one. Thsi is a really fabulous CPU, as it did 500x9 with 1.63 without any problems. We had really high hopes with this one, unfortunately we hit a FSB wall at 505Mhz, which we couldn't get over even with 1,8v on the core. Still, at 4500Mhz - 1,63v @ -62C it is a very stable and cold E6600.

The last candidate was a Prescott core Celeron D346 (3.06Ghz, 522FSB, 256Kb L2 cache, 90nm). We got into windows at 4920 a few times, but when we started to tweak the Super Pi Score we couldn't get it to go higher than 4,83. We were impressed with this little chip, even though we wished we had an 65nm Cedar Mill core with 512kB L2 :) Result : 31.938s

In the end, on behalf of M1cutzu, we played with X6800 + EVGA 8800GTX, but this was the hottest CPU I have ever seen, so it really lacked the appetite for glory (at least under Dry Ice), hitting only 4100Mhz 3DMark 2001 stable. So we just had a run of 3DMark 2001, just to finish the Dry Ice, and we hit a modest score: 73682 points. Well, it's small, but it is ours :D

In the end we would like to thank again to M1cutzu, Sang and Xwoodoo for all the help they gave us, we really couldn't do it without them. We would laso like to thank Iubitel, he was there with us all the time, giving a helping hand.

We have learned a lot from this first Dry Ice session, the temperatures and clocks we played with really opened us an appetite for this type of benching, which we will repeat again really soon :). On the other hand, we learned from the support we got from the big guys what it means to really be involved in the development of the enthusiast community here.

I will leave you with some pics now :)
Monstru
08-07-2007, 01:16 AM
And some more pictures :)

1. The setup
2. Xwoodoo giving a helping hand
3. Iubitel and Matose
4. Yours truly
5. Matose doing some hard thinking
PytonOrm
08-07-2007, 01:29 AM
haha :D cpu-z says 474 x 6 = 2845mhz and superpi at 10.687 s ??? wtf dude :) whats wrong and good job anyway m8
Monstru
08-07-2007, 01:42 AM
haha :D cpu-z says 474 x 6 = 2845mhz and superpi at 10.687 s ??? wtf dude :) whats wrong and good job anyway m8

I forgot to mention that we used C1E active, to get temps as low as possible. We would boot into windows at 4500Mhz, use setfsb to get at 4740Mhz, let the CPU cool down (the multi and vcore would come down allowing the chip to cool) and then we would run SuperPi. Thats why Cpu Z shows 2845, but you can see from the fsb the real clock 474x10= 4740Mhz. Here is one run without C1E active (we couldn't get over this if we wouldn't have used C1E) :)
Solarfall
08-07-2007, 02:04 AM
I forgot to mention that we used C1E active, to get temps as low as possible. We would boot into windows at 4500Mhz, use setfsb to get at 4740Mhz, let the CPU cool down (the multi and vcore would come down allowing the chip to cool) and then we would run SuperPi. Thats why Cpu Z shows 2845, but you can see from the fsb the real clock 474x10= 4740Mhz. Here is one run without C1E active (we couldn't get over this if we wouldn't have used C1E) :)

hey nice trick have to try that with my e6850 later on this week.. no superPi m32 scores ??
btw great scores Monstru :toast:
Cupcake
08-07-2007, 02:06 AM
Pretty nice job :D

you forgot to disable the 6x multi while idle thingy :p oh well
Monstru
08-07-2007, 02:16 AM
Solarfall: thx m8 :) I don't think it would have been Spi32M stable, because of the heat :) As for the trick, it helps if you are at the limit with the temperature, just a little bit, but it helps.Anyways, it is "Matose copyrighted" :D. You have to let it cool, and then run SPi....then let it cool some more (until you get the lowest temp possible - in our case -52C) then runs spi again. And so on.

Cupcake: I explained why we left that thingy active ;)
Morais
08-07-2007, 06:50 AM
Good job bud, keep pushing more!
Monstru
08-07-2007, 07:08 AM
Thx man, we will try for sure. We just need some time and a good chip, and we can have even more fun :d
eva2000
08-07-2007, 07:21 AM
nice congrats..

what were you applying with that brush on the actual cpu socket pins ? dielectric grease or nail polish ?
Monstru
08-07-2007, 07:24 AM
Thank you!

We were applying nail polish on the caps inside the socket.
Florent
08-07-2007, 06:24 PM
Nice Overclock :up:
eva2000
08-07-2007, 06:33 PM
Thank you!

We were applying nail polish.
how effective would nail polish be against corrosion of the socket pins ??
Micutzu
08-07-2007, 09:39 PM
They're insulating the caps in the middle of the socket with nail polish ;).
dinos22
08-07-2007, 09:58 PM
it would be hard to cover CPU pins but middle socket is a neat idea

nice results there guys

i hope to do some benching myself with my E6700 very soon :up: on dice
Stapler
08-07-2007, 10:15 PM
How did you use the dry ice? If you crush it into ice and make a sort of slushy with the acetone you'll be able to pull better load temps. You'll need to constantly tamp the mix to keep gas pockets from forming, but with a good pot and a good slush you can keep the temps from rising at all, even through 32m. Still the first time is always a little difficult, very good effort and the times are pretty impressive for the clocks as well.
dinos22
08-07-2007, 10:20 PM
dice pellets in sydney are very small...at least ona few occasions we used it......they were like large beans basically and tube was holding the temps nicely but maybe these all alu tubes are weak at the bottom donno
Xevipiu
08-07-2007, 10:41 PM
E6700 - This one was very hot, and our Alu tube was pretty thin, so we couldn't get it higher that 4740Mhz (Spi Stable), vcore 1,71v, @ -52C. Still, we consider this to be an ok result for starters, with a CPU that doesn't go higher than 5,1 on LN2... Result: 10.687s


1.71v full?

Nice O/C ;)
Monstru
08-07-2007, 10:44 PM
Yeap, we were insulating the caps in the middle of the socket, as M1cutzu tought us :D

Stapler: we used dry ice pellets with alcohool, and it worked pretty good for us, being the first time and all :)

dinos22: our pellets were way smaller than large beans...they were just a little bigger than rice.

xevipiu: I suppose that there is a little drop there, but considering the board, I would say that the vcore was pretty close to 1,71v.
dinos22
08-07-2007, 10:51 PM
Yeap, we were insulating the caps in the middle of the socket, as M1cutzu tought us :D

Stapler: we used dry ice pellets with alcohool, and it worked pretty good for us, being the first time and all :)

dinos22: our pellets were way smaller than large beans...they were just a little bigger than rice.

xevipiu: I suppose that there is a little drop there, but considering the board, I would say that the vcore was pretty close to 1,71v.
how large are your beans there dude :D haha

those are small pellets so it seems you need a bullet tube from Duniek or one of the other guys here

or a solid one i guess
Xevipiu
08-07-2007, 11:21 PM
On my P5K Vanilla it was 1.58/1.62v full (no-mods) :( 4.820 DI (http://img62.imageshack.us/my.php?image=2m482dy6.jpg) And on 5.118 LN2 (http://img57.imageshack.us/img57/6489/1mtopsensemodsix0.jpg) 5.3 1.74v (562fsbwall x8)

With time it tries with +vcore, out the results ;)
Monstru
08-07-2007, 11:44 PM
Xevipiu: nice clock you got there for Dice :) We had to find a vcore which would allow us to get the clock we wanted without makeing it too hot. So, that was the good vcore for us. Everything above that would crash even at a lower clock.

dinos22 : Yeap, we have small pellets, as you can see from the pic. Duniek's pots sounds good....
dinos22
08-07-2007, 11:53 PM
wow i wouldn't even call that pellets man

that's more like dust :D
Monstru
08-07-2007, 11:57 PM
Now now, the picture is not quite good quality :)
Cupcake
08-08-2007, 12:26 AM
Cupcake: I explained why we left that thingy active ;)

I only look at pictures ;)
matose
08-21-2007, 09:09 AM
Thx for the positive feedback. This is only the beginning :p: !
BeardyMan
08-21-2007, 09:20 AM
wow i wouldn't even call that pellets man

that's more like dust :D

dust is needed for good dryice session :D

you can press it and improve temps, and no air between de ice and pot is a must...with dust :D
Monstru
08-21-2007, 09:33 AM
Thank you for your suggestion, pretty soon we will try to get some E6850 action and we will try to refine the whole process.