Folding Forum

Memtest works like the FAH client and support -gpu X flag

Run it from a command line window :

MemtestG80.exe <amountofmemorytotest> <numberofiterations> -gpu X

i'm running vista ult 64
i hit the windows button(search), and type command, then enter
the Administrator: Command Propmpt bos appears with C:\Users\computername
i type MemtestG80 enter
i type MemtestG80.exe enter
i even tried just typing the <> stuff

it all gives me memtestg80 in not recognized as an internal,,,,,......yada,yada,yada

i'm definitely not a command prompt guru,

any help in what i'm doin' wrong?

tx

btw, on it's normal pass when i double click the exe in it's folder it passes 50 iterations with no errors..??..??.....

M$

you need to be in the folder where memtestg80 is.. You'll need to use the cd: command to change into the right folder.

HA!!!!
i'm in!!! this is embarrasing, i should've paid more attention in 'dos' class! what a noob
after some fiddling, i finally was able to test each card at 384mb 50 iter's ramclock 900 core clock 600.....
both cards, no errors!!!

now can we kill these or at least stop sending them to me???

now, can i set the shader's up and do the same thing???

M$

OK, well they both passed up at 1700 on the shaders as well!!!!!

M$

yup....... NOOB ALERT DANGER DANGER NOOB ALERT DANGER............

Nah glad you figured it out..

LOL!!!!!
yep i figured it out and now it says what i thought was the case originally!
It's not my equipment or my OC or my not OC, It's the work units!!!!!

Soooooo, let's stop sending them to Mr. Scary's Machine!!!!!!

?????


M$

It's not the WU as they don't break my 8800GT or your 9800GT (basically the same cards). I think it's your motherboard. Given how few ppd you're getting, why not try taking one of them out? Then try with just one and see if it folds OK. If it does, it's your motherboard's support for two cards which is faulty or your PSU just can't handle two fully loaded cards. The WU you don't like really push the cards quite hard, I believe. And that draws more juice.

thanks for the response, Halo.
As for the mobo, its fold with 3 & 4 cards before.
As for the psu, i believe it's either 650 or 750W's, plenty of power for 2 8800GT's. this same psu had powered 2 gtx260's, also, i've got a psu tester and all rail(s)/outputs are good to go.
When i get another one of the wu's, i'll pull one of the cards and test. The underlying reality is both cards and the mobo will fold everything else under the sun, OC'd as well, with NO probs. Soooo,
that truly points to the WU's.

Either way with PG's technology/server's i can believe they can't direct those wu's somewhere else besides my rig! If they keep coming back and eue'ing, stop pushing them my way!!!??!!!

M$

If it folded with 3 or 4 cards before, have you tried moving the cards from the current PCIE slots into the other two. Multiple PCIE lanes are a real PITA in most motherboards and 680i was a fairly early attempt to do more than two.

I think that's probably my next step/option. i really don't care where they wind up on the mobo, just so long as they fold. this is gonna be liquid cooled so after that's setup, i can't really be flopping the gpu's around, LOL!!! I'll keep you guys posted, and again, thanks all for the responses and suggs!

M$

Either that, or reduce the overclocks temporarily on the affected system and see if self-tests start passing.

Also remember to set up your GPU fans to keep all the cores under 80C. I use RivaTuner for that. Despite what Nvidia says, many later cards from the 8800 model generation have been erroring out if not failing after prolonged exposure to ~85C folding temperatures.

Thanks Wrish, the gpu's aren't OC'd, although they passed memtestG80 at the higher shader settings(1700&1750).
Do you mean any cpu OC as well?
As for the fans, they are always at 100%, LOL!!! The temps are usually in the 60*C area, sometimes into the low 70's but not to much. I'd rather have the cooler temps and the noise then, well, you know, probs and quiet!!!!! They are actually idling right now at a nice and cool, 42*C! BRRRRRRR!!! Relatively speaking!

thanks again, for the response(s)

M$

When troubleshooting, try all the possible culprits, so yes, disable the CPU OC so you can rule it out as a problem. The GPUs pass OCCT-GPU or Furmark as well, right? I'm skeptical of MemtestG80 detecting all the problems because Memtest stresses memory, not shaders.

You can also copy the folder containing the buggy work unit and test it on the GPUs without waiting to be assigned one. You'd be using the command line (systray client works fine in cmd.exe) and modifying the configuration to set the workload on -gpu 0 or -gpu 1.