Gautam has good advice, try to keep the rig under some kind of load, that will minimize problems with the board. As soon as the caps freeze over Vcore will go wild and you have to quit, that won't happen in most cases if you keep the rig loaded most of the time and not too cold. Most chips won't need anything under -140 to rock anyway so explore the -120 to -130 ranges before going colder.
This also applies to the cards which may bug under -100 or even warmer. It's actually more important to keep the card loaded so the PCIe slot doesn't freeze over ending your session, it happens alot faster on a card than the mobo. Be sure to grease the PCIe slot btw, thats something you can't forget.
These are my dewars. Not pictures of MY dewars, but they are both MVE Cryogenic Dewars Model AL-30, and this is a pic of that model, and this is what my dewars look like.
I cant find the volume of these dewars though....I assume that model AL-30 means 30Liters?
Also, speaking of LN2 on GPUs, I have not read about greasing the PCIe slot. How do I accomplish that?
What do I need to do regarding starting up with my GPU when Im benching it with LN2? Dry pot to start as well? Boot to windows...Light OC then mild LN2 then start to add more OC and more LN2 incrementally?
On the cards here's what I do. Because they are usually running high volts you need to have them somewhat cold maybe -30 to -40 idle and drop the temperature going into a bench. -80 will suffice in most cases loaded. I don't ever run a volt modded card on a dry contanier, that would be risky imo. Most GTX's for instance will run 900MHz @ -45 with the proper voltage. ATI is a little different, I have yet to freeze my 2900's but I think they will need more cooling because they run soo damn hot. You just have to test things with your hardware. Btw one idea you may want to consider. In order to maximize your benching success you might want to bench the card on LN2 with the CPU on air/H2O, that way you know exactly how cold, how much VGPU and the clocks it will run. If you try and figure this out at the same time your figuring out how to run the processor chances are you'll wish you'd done it seperately
After tonights lack of success I would have to say I agree with you.
Ill probobly do LN2 on the CPU tomorrow night with one of my basic GPUs on stock cooling.
Next night if my card is dry Ill do 8800GTX on LN2 with cpu on air.
After I get the hang of each individually THEN Ill go for a 2 pot session.
I think that is a MUCH better plan. Im glad my card was still sticky tonight in hindsight.
Also, where can I read about greasing a PCIe slot? I havnt read that before. Di-electric grease? Any suggestions on where to obtain? How much to apply and where?
Thanks again
Aint it great having a new guy around to badger you guys nonstop for all of this info
you can get dielectric grease at any auto parts store, it's what gets put on spark plugs and as far as applying, just put a layer across the finger of the card on both sides and slide it into the pci-e slot, to bad about the condensation, I'm sure I'll be the same on my first ln2 run
Once again, the key is keeping the card loaded. The 7600GS that JC posted up obviously has a very tiny load. (That's why I killed 3 of them running on DI, never even got two of them to POST )
The 8800 should be okay. kp never uses any grease...but from what he's told me, if he sits idle for too long, he does get frost on the fingers as shown in JC's pic. For your first few times consequently you should put some grease in the slot, but other than that his insulation guide works wonders. That's exactly the insulation method he uses, and no one has benched the 8800's more than he has.
A GTX will freeze over pretty quickly if it's not loaded, I'd grease it, just a small bead on the slot. Make sure the SMD's nearest the slot are covered also if you didn't get them with the silicone spray. If the card is coated properly you don't need any insulation on the back of the card.
A GTX will freeze over pretty quickly if it's not loaded, I'd grease it, just a small bead on the slot. Make sure the SMD's nearest the slot are covered also if you didn't get them with the silicone spray. If the card is coated properly you don't need any insulation on the back of the card.
All-in-all it went very very well. NO repeats of last night. All the hardware was in fine working order. I had NO coldbug, and in retrospect and with tonights testings in mind, I can say with absolute certainty that I coldbugged/coldbooted whatever you would like to call it yesterday.
In summary, I spent the entire night on LN2 (4 hours tonight). Overclocked a great amount, and if I wasnt allready OCing a chip that I beat to death on dice previously, would likely have some very very good screenies. Unfortunately, I was unable to gain any additional fsb or speed with LN2 as compared to dice.
Question: Is this possible that the chip maxed out at whatever -75C on dice could do for it? Or....is this more just a need for me to learn/experiment more with this chip on LN2?
In more detail, I did a fair amount of testing and kept some decent notes to look back on and try to figure out the sweet spot of this chip. Please tell me if my assessment is correct or off base.
Dry pot BIOS boot 31C. After 3-4 mins 36C. Not great but not that bad and I wasnt about to remount so I dropped the LN2 in until abou -60C. Booted into windows at 9X450 with no issues at highish volts. At around 85ishC I went to 542 FSB X 9 but didnt screenie since I have almost exactly 542 on dice allready. Jumped to 545 and froze (possible reboot too notes dont say).
Increase Vcore nd Vmch a tad. Posted ok at -92 but a bit slowish/choppy. Booted into windows at 475X9 again at -62C no probs. SetFSB froze when applying 540FSB at -95C but there was no reboot.
Increase Vcore and Vmch again. Booted at 450 X 9 this time. Got 542 FSB again at -90C but when applied 545 FSB at -89C computer froze with no reboot.
Maxed out all volts (just like I did with my best OCs with this chip on dice).
Boots from now on are 475 X 9 and went without any issues every time at anywhere from -60C to -90C.
SetFSB froze applying 540FSB at -105C. .Tried colder here but didnt pan out.
Thought I was able to get the 545FSB screenie at -91C, but I guess it froze a millisecond before I got to save it Was able to successfully apply 545 FSB at -91C though!
SetFSB froze at 530FSB at -92C with the pot getting colder
SetFSB froze at applying 525 FSB at -112C
Windows froze without any provocation at -125C, and I did not go near -125C or colder than -125C again tonight.
So......from my assessment it appear that additional cooling over dice offered me nothing for this chip/mobo combination. Sweet spot appeared to be approx -85C t perhaps -91C. Said sweet spot on LN2 did not offer me any better overclocking capabilities than what I allready received on dice, seen here.
Agree or Disagree?
Thanks again guys
8800 GTX was dry this evening, so tomorrow night Im gonna hit up my 1st practice run 3D on 8800GTX LN2, but will switch the CPU from the Dragon Evo to my TRUE 120 to focus on learning the Tek9 pot. TRUE 120 is benchable at 4050 on my P5K premium, so actually I may get 4100 or 4150 perhaps on the Mars even. Not all that far off from what I would likely get on dice/LN2 anyhow actually unless someone sees a flaw in my logic or technique.
I would try less vcore for starters, around 1.7V. It does appear that your chip bugs @ -125 on this board. I know increasing CPU Reference pushed the coldbug lower on my 3070 by quite a bit, might be worth trying out on this setup. If you didn't lower PL to say 12 or 13 before using SetFSB, you should try again. There is a good chance however that better cooling won't do alot on this chip, it may be FSB limited.
Good work doc, a few more sessions like this and you'll have a much better feel for things
Yup, I took this pic 20min ago. You gotta try this stuff, it gives you peace of mind as well as more space between the cards. This is just DI @ 1.413V around 1050MHz core.
Less Vcore huh? From my limited knowledge it seems counterintutive but Ill give it a try and see what happens
CPU Reference.....what is that?
Lower the PL? Are you speaking of performance level again on memset, or PLL volts?
I hardly touched memset tonight. That screenie is from my dice session the other night. I just showed it as my highest CPU OC which happened to occur on dice.
Since I was working ONLY CPU-Z and CPU OC tonight, I didnt touch the ram. I set it at stock timings (5-5-4-12 for the ram I was working with tonight) and lowest mhz allowed.
Should I be messing with memset when exclusively working on max CPU-Z shots?
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.