PDA

View Full Version : First DI-Run......big problems !!



killingspreez
12-18-2005, 06:53 AM
Hi,
this we we did the first DI runs with our systems.
we had an opteron 144 and my 3700+ SD, 2*dfi nf4 (ultra-)D, ocz 520watt
g.skill GH / corasir xms 3500 and so on.....

well now everything was okay.....
the insulation:
(under the armaflex was a layer of plastik-silicon-spray)
http://img464.imageshack.us/img464/6390/imgp00187xs.jpg

everything was okay...no condensation or ice even after several hours of benching.

99,9x% aceton + the DI did a great job, too. (t1= inside of the tube)
http://www.tigersclaw.de/work/tribal//DSCN2621.JPG

everything ready for the first try:
http://img464.imageshack.us/img464/8085/imgp00200fa.jpg

we first booted into win (default clocks) and then added DI to the aceton to drop the temp in a gently way. so after a while we had this temps:
(t1= inside the / t2= tubeIHS of cpu )
http://img464.imageshack.us/img464/7268/imgp00263ag.jpg

in windows it looked like this:
(100% stable)
http://img236.imageshack.us/img236/9798/3700cold7sg.jpg

so we thought the 3700+ would not have the cold bug......
we tried to overclock the cpu and it did well untill we hit 2,9ghz (2,93 ghz is table on air, so we were not impressed).
super pi failed....so we raised v-core and so on but no matter how much v-core we gave this sucker it wouldn't get stable above 2,9ghz (even 1,85v didn't help). we had still the same low temps......

well then we raised the temp to search for a certain temp where the cpu would be stable. but no luck we tried everything from -77°C to +10°C everything was unstable above 2,9 ghz....
we also tested ALL dividers for the ram and different ram-types/gpus....nothing helped.

now what was the problem ? is this a cold bug or not? we tested both boards and no matter which board we used the 3700+ didn't go above 2,9ghz (on air even 30mhz more and stable!).

we also tested the opti 144. it had the classic coldbug at around -8°C (windows freezes and it didn't boot until the temp was above -8°C) and it hit 3,15ghz (3ghz was possible on air).

so what went wrong? why the hell did this 3700+ nothing above 2,9ghz???
does it have another type of coldbug or did we do a mistake?

thx for any help.

mfg

EDIT: further pics of dothan-sys we follow

EDIT²: we also had a ram-cooling device but we didn't run it cuz we were afraid of damaging the mobo (or ram) caused by the very bad insulation :D
http://img529.imageshack.us/img529/4199/dscn26128cx.jpg

Jort
12-18-2005, 07:44 AM
is it so hard to resize pics.....

how the hell should we have a good look at them?

btw first things don't run as smooth as the second run ;)

good luck

and btw NICE memcooler:D

[XC] moddolicous
12-18-2005, 08:27 AM
Yea, those pics are gigantic. I hope that mem in that mem cooler is BH-5. Did you make sure there was good contact with the IHS?

killingspreez
12-18-2005, 08:38 AM
the contact was not the problem...it was very very cold (look at the pics+describiton) ihs of the cpu was -60°C and the cpu-die was about -45°C
so yeah contact was very good.

has anyone seen an issue like this before?
how to fix it? i'm still waiting ;)

mfg

afireinside
12-18-2005, 09:47 AM
Contact is ALWAYS my issue when that happens.

K404
12-18-2005, 10:15 AM
Well...even bad contact should still mean the core is in contact with something thats subzero so there should be some kind of gain. Loaded temperature gains would suck, but there should be some improvement...shouldnt there?

afireinside
12-18-2005, 10:26 AM
Not really... I had perfectly fine 30Cs loaded once and could run 2750mhz 1m but 3dmark crashed at stock. Reseated block and got same temps but 2.8 3d stable.

[XC] moddolicous
12-18-2005, 10:40 AM
Did u put vaseline in the socket? I think your supposed to do that to prevent the pins from getting condensation.

Flib
12-18-2005, 10:49 AM
We didn't put vaseline into the socket but instead we used some plastic spray.
As he already mentioned we tested 2 Boards with 2 CPU's.
The Opteron acted quite normal, but his SanDiego got a strange "Coldbug".

There were no problems with condensation. The Opteron worked well expect the Coldbug. We could only bench @ -10C. Contact was good, all was fine. We reseated the tube at leat 5 times. It has to be the CPU.
This SanDiego has symptoms, we've never seen.

gclg2000
12-18-2005, 10:55 AM
cool memeroy cooler

^don.k's^
12-18-2005, 11:53 AM
Isn't a no-boot-coldbug but is cold bug, i bet you couldn't boot @2,8ghz, even @2,7 i'd say you couldn't also... Run prime on background or let the dice time to evaporate and let the temps raise, hold the temps about 230-240 (smartguardian measure) and try better clocks, that what i had to do with both my 144 and 146 opty's, i hope i can help you... :toast:

killingspreez
12-18-2005, 12:43 PM
hi,
it was bootable and at least 4m stable at 2,9ghz with low v-core but 20mhz more (no matter how much v-core) it crashed.

i think maybe its the missing pin....but i'm not sure.....could this missing pin be the aswer to all this trouble?
i'll show you the exact pin if you can give me a pin-pic or something like that.

mfg

hm i'm not too sure...mabye there is no pin but i mean here:
http://img231.imageshack.us/img231/9816/9390xs.th.jpg (http://img231.imageshack.us/my.php?image=9390xs.jpg)
one in the lower left corner is missing but i dunno maybe this is normal.

mfg

last edit:
okay this is quit normal there is no pin in this corner so....there must be another fix or reason for this bug :(

Tim
12-18-2005, 02:23 PM
Dude resize the Pics ASAP!!!! :slapass:

I'm not even gonna look at it this way..... :(

afireinside
12-18-2005, 02:41 PM
Did u put vaseline in the socket? I think your supposed to do that to prevent the pins from getting condensation.

Only for 24/7. For DIce/ln2 greasing the socket is an un needed mess.

lawrywild
12-18-2005, 03:22 PM
http://img464.imageshack.us/my.php?image=imgp00187xs.jpg

http://img464.imageshack.us/my.php?image=imgp00200fa.jpg

http://img464.imageshack.us/my.php?image=imgp00263ag.jpg


erm, contact I reckon as stated above...

[XC] moddolicous
12-18-2005, 03:47 PM
Only for 24/7. For DIce/ln2 greasing the socket is an un needed mess.
Alright, thanks. Didnt know that.

T_M
12-18-2005, 06:51 PM
Can you give a full BIOS settings list for CPU/RAM for what you were trying to get teh SD working.

Dumo
12-18-2005, 10:26 PM
Try with ACPI multiproc. It will help a lot for your 3700 or Opty.

Tribal84
12-19-2005, 12:38 AM
same things with the other config ..
p4p800se + dothan 730 = @ air 2650 MHZ @ di ca. 2900 MHZ .. i want more :)

but for the first di run, it was o lot of fun and we learn a lot to.

next session i will use my ramcooler with better iso :)

marvt74
12-19-2005, 03:39 AM
To test the SD next time keep trying different levels of acetone to vary the temps.
SO it's not always at stupid cold level

Tribal84
12-19-2005, 03:43 AM
we tested the opteron + 3700 with -50 to + 15 degress and they do strange things ...

Eldonko
12-19-2005, 12:34 PM
My 3700+ is the same. There is no cold bug at all under single stage phase (-50) but with dice I cant clock the think past 2800Mhz without it freezing.

To work around it I had to run prime in the background at priority 1 during my PI runs which is far from ideal when shooting for times. As you can see here the 1M time is super slow for >3200Mhz. I can run faster than that on phase. =/

http://members.shaw.ca/zryan/dice2-7.JPG

The only way to beat cold bug is to do what Dumo said or run apps in the background. Cold bug is the suck. :mad: :mad:

killingspreez
12-19-2005, 01:02 PM
i think you don't really understand our problem.... :(

here the facts:
1) no stabilityissues from +15 to -79°C (fully stable if core is under 2,9ghz)
2) over 2,9ghz no go (no matter what temp or v-core)

so we don't have to fight against the COLD....it's something with the clock that causes this problems. it could be the absolute maximum for the cpu but this isn't possible because on air it is stable at 2,93ghz.....so there was no gain from cooling with DI.

now i wanted to know how i could work around this "bug" for the next session.

thanks ;)

mfg

Eldonko
12-19-2005, 01:35 PM
I do understand and my post above still applies. the COLD is causing the chip to stay at 2800Mhz. Boot at 2700, run sp2004 in the background at priority 1 and increase clocks with clockgen. You will see you can exceed 2800Mhz with the added heat. Cold bug works in various ways, sometimes it limits RAM clocks, sometimes HTT, sometimes Freq.

My 3700+ will not boot over 2750 under dice and freezes at any speed above that. Add heat and it goes above 3200Mhz.

killingspreez
12-19-2005, 02:11 PM
okay
now that's something helpfull! *thumbsup*

any other suggestions?

mfg

Flib
12-19-2005, 02:16 PM
Finally after 26 post we got an answer :woot:
Nobody else had such problems?

Eldonko
12-19-2005, 02:30 PM
I tried many things to defeat cold bug, but there is no way to totally get rid of it. Basically if your CPU is cold bugged, you have to work around it. I thought the expert board and 15 levels of drive strength would help but it does no better than Ultra/regular SLI. I tweeked for over a day with Expert and there is nothing you can do other that add a little more heat to the mix. Here are the ways to work around cold bug:


Monitor temps closely and add very little dice at a time in lots of acetone. This way you can carefully tweek the temps to levels in which the cpu works properly.

As Dumo said, use ACPI multiproc. I haven’t tried this yet, but will next dice session.

Run SP2004 at prio 1 in background.


None of these solutions are ideal, but it is better than nothing. :) Good luck.

[XC] leviathan18
12-19-2005, 03:38 PM
try to run ram on dividers to see if you are putting to much preassure in the memory controller

Optimus Prime
12-19-2005, 06:13 PM
We didn't put vaseline into the socket but instead we used some plastic spray.
As he already mentioned we tested 2 Boards with 2 CPU's.
The Opteron acted quite normal, but his SanDiego got a strange "Coldbug".
You didn`t mention what tupe of "plastic spray" you used,but i think this might be one part of the problem.When i was having like my 1st DryICE run i also applied some no-name plastic spray in socket and this made my o/c totall crap.After that i cleaned every remain of it,and than everything went well.Maybe you are having same problem,but you can never be 100% sure if you don't test your CPU without this spray.Of coure,you might be dealing "only" with cold bug,but nevertheless try ACPI multiprocessor tweak and see if that will help you pass 2.9Ghz.
Good luck ;)

T_M
12-19-2005, 07:18 PM
Finally after 26 post we got an answer :woot:
Nobody else had such problems?

Well, if you'd answered my question (BIOS settings) we'd get there a lot sooner

Flib
12-20-2005, 02:14 AM
Can't say the exact Bios setting.
Maybe Killingspreez can do later.
But we tried everything to get this SD stable.

I don't think it's the spray, cuz the Opteron worked well (expect the early coldbug)
It doesn't matter which settings we chose, we couldn'T oc higher than on air.

Tribal84
12-20-2005, 04:05 AM
WE used Plastik 70 Spray ..
I dont think it´s this spray the dothan worked well ...