• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Determining GPU Core

C

capreppy

[H]ard|DCer of the Month - April 2009
Joined
Jul 4, 2007
Messages
3,410
Running 3x9800GX2 (from Kendrak). I think I have heat problems. I'm running them inside a NorcoTek RPC-470. It is a 4U Server Case.

MoBo: MSI K9A2
CPU: AMD PII X4 940 Black Edition (stock HSF & stock Voltage, but OC'd to 3.4 using multipliers). Temps are 49 which is well within reason.

I consistently have EUE's on GPU 2 & GPU 5. My belief is that this is the middle 9800GX2.

9800GX2 #1 = GPU 1 & GPU 4 (correlates to -gpu 0 & -gpu 3 flags)
9800GX2 #2 = GPU 2 & GPU 5 (correlates to -gpu 1 & -gpu 4 flags)
9800GX3 #3 = GPU 3 & GPU 6 (correlates to -gpu 2 & -gpu 5 flags)

Can someone please confirm?

I'm currently using eVGA Precision 1.5.1 to set fans and OC's (well currently well underclocked because of the EUE's). 1.5.1 addresses up to 4 GPUs. Precision is much simpler to set then RivaTuner.

Should I bite the bullet and use RivaTuner (latest version, but can't remember the version number) to set OC's on all 3 9800GX2's. I could downclock the middle card so that it doesn't generate as much heat and leave the outer 9800GX2's at stock (or minor OC). I don't have an issue with running the fans at 100%. I have already downclocked the Core & Memory to reduce overall heat, but that doesn't seem to help as much as I thought it would.

I have created a little space (not much since they are sandwiched together) between the cards. In addition, there are 4x80mm fans (server type as in really loud and push a lot of air) in front of the cards to provide cool air from the front of the server.

Anything else I could do? I've thought about pulling the middle 9800GX2 and sticking it in a different box and leaving the middle slot open. That may be what I need to do, but am trying to avoid that. Reason is I am trying to maximize PpD
 
Well, if you want to confirm which card it is, try pulling it out of the machine and monitor your other clients for EUEs.

If it is in fact that card, then you should try lowering the clocks and increasing the fan speed to 100%, as you already said. More than that, there really isn't anything. There is also the possibility that the EUEs are being caused by something else, and that the card just isn't playing nice with the rest of the box, but there's no way to really know without playing around and doing some troubleshooting.
 
I know that there was a bad batch of WU, but I thought they were gone. In looking at the temps via Precision, my temps aren't all that bad (in the high 50's, which I assume is ok).

It seems to run ok at night (lower ambient temps) so for now I'll just run at night. I can't really debug it much until the week after next. I may just move the middle one to a different box.
 
cap your cards aren't going over 50c or so? right now mine are sitting at 65-70c and are slightly shader overclocked and I can't remember the last time i had an eue. So it seems like you should be good to go the only time i've had an eue lately was when i forgot to extend the desktop.
 
I am currently running 178.24. I think remember trying 182.08 (I think I had some issues with them). What driver are you using?
 
i'm currently using 181.22 on that machine i have 2 9800 gx2's and 1 9800 gtx on the same mobo you are using.
 
Mine have always loaded in the order the drivers load in Windoze.

9800GX2 #1 = GPU 1 & GPU 2 (correlates to -gpu 0 & -gpu 1 flags)
9800GX2 #2 = GPU 3 & GPU 4 (correlates to -gpu 2 & -gpu 3 flags)
9800GX3 #3 = GPU 5 & GPU 6 (correlates to -gpu 4 & -gpu 5 flags)
Etc .....

If you have two monitors then plug them both in and use Disply Properties -> Settings -> Identify to see which monitor belongs to which card output.
Move the second monitor to each output as you use the first monitor to see the identify button.
Cuda and the -gpu flags follows this setting minus one.
Primary Display 1 = -gpu 0.
Secondary Display 2 = -gpu 1.
Secondary Display 3 = -gpu 2.
Etc ........

As thats the order Cuda sees the cards.

Luck ............ :D
 
thats a good idea on the monitor! mine has always just installed like you showed -0-1 being card one etc..
 
Hmm. That throws my current theory right out of the window :(

Why would I consistently get EUE's on 2 & 5 then? They are on separate cards. Temps according to precision on GPU2 are actually ok
 
do you occasionally get those cards working? or do they only eue?

did you install the cards all at once or one at a time? it might work to uninstall all the cards and just one at a time bring them up to make sure they work then install the next one? Also i suppose you could lose a gpu on 2 different cards? seems unlikely but who knows...
 
Are you running these overclocked? Your temps seems fine, so wondering if your OC is just too much.
 
84gthatch said:
do you occasionally get those cards working? or do they only eue?

did you install the cards all at once or one at a time? it might work to uninstall all the cards and just one at a time bring them up to make sure they work then install the next one? Also i suppose you could lose a gpu on 2 different cards? seems unlikely but who knows...
Click to expand...

These sem to work during the night when ambient temperatures are lower.

PTRMAN said:
Are you running these overclocked? Your temps seems fine, so wondering if your OC is just too much.
Click to expand...

These were actually downclocked significantly due to EUE's and well the downclock didn't work.



I had some time this morning and moved the middle GX2 to a different box. Seems to be much happier. I may need to limit myself to 2 x Dual GPU Cards until I can figure something out. Short of WC'ing, I can't really think of any way to keep temps down on the middle card. I have to stay in a Rack configuration so my options are severely limited.
 
mine didn't eue when they spiked into the 90c range for awhile because i forgot to turn the fan on them -_- it's weird that yours eue so much when they don't seem to be getting that hot? what I've done since is put a folded piece of paper between all the cards which seems to help and also put a 120 mm fan pulling the hot air out of that area so who knows that might help? you could also put a dual slot single gpu card in the center those cards don't seem to have as much issue venting the hot air outside so the slot is at least not fully wasted?
 
You must log in or register to reply here.
Back
Top