Im having a big problem with this GTX 295
http://setiathome.berkeley.edu/show_host_detail.php?hostid=5253344
The thing is no matter what OS or driver I use, or even how far I underclock the memory or the gpu, the card uses like 514mb of memory each GPU, and produce a lot of errors, and like 2x at day, artifacts and hang up the entire machine.
I try to RMA it, but it seems if the card works in games (right) they won´t trade it :(
Does any one have a clue on what is happening?

---

Im having a big problem with this GTX 295
http://setiathome.berkeley.edu/show_host_detail.php?hostid=5253344
The thing is no matter what OS or driver I use, or even how far I underclock the memory or the gpu, the card uses like 514mb of memory each GPU, and produce a lot of errors, and like 2x at day, artifacts and hang up the entire machine.
I try to RMA it, but it seems if the card works in games (right) they won´t trade it :(
Does any one have a clue on what is happening?

Nope, but mine does the exact same thing. I'm replacing it with two GTX 480's.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

It produces several kind of errors, it seems. I have no clue on what they mean either :(

---

I just got off the phone with Gigabyte about the two 295's I RMA'd due to overheating and blackscreens, along with the same errors mentioned above. They sent me an email saying the cards were unrepairable, but that there was no replacement stock available.

They suggested in the email that I should call their customer service and tell them what model of card I would like to replace the two 295's. Long story short...they wanted me to take some refurbed ATI cards. I (not so politely) declined their offer, so they are going to refund my original purchase price.

I know, probably not 100% on topic, but had to let off some steam.....

I noticed a great deal of -9 errors that occur within 10 seconds of starting a WU. This usually means theirs something wrong with the GPU. I wouldnt be surprised that running seti stresses the GPU into errors.

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I noticed a great deal of -9 errors that occur within 10 seconds of starting a WU. This usually means theirs something wrong with the GPU. I wouldnt be surprised that running seti stresses the GPU into errors.

Yes, they happens after a first crashed unit and if im not around to stop it, the card will ruin all my cache producing that 0.01 credit errored units :(

---

Wondering if that can´t be fixed, can I use a cc_config file that let me ignore GPU 0 or GPU 1? But just in case that there is not a better workaround this, because that will let me with a one GPU 295....

---

I dont think a cc_config is going to fix a hardware problem. This appears to be beyond the BOINC barrier.

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

That memory usage is far beyond what my other 295 uses. Besides setting video performance to performance on Win 7, what else can I do to minimize memory usage?

---

by better video I assume you mean you've turned Windows 7 aero off?

I'd swap the card out onto one of your systems that is also running the same cards. does it continue to produce bad results?

using over 500mb memory? are you saying the card is using 500Mb of system ram?

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

Wondering if that can´t be fixed, can I use a cc_config file that let me ignore GPU 0 or GPU 1? But just in case that there is not a better workaround this, because that will let me with a one GPU 295....

I don't know if that would fix it, but there's the <ignore_cuda_dev>N</ignore_cuda_dev> option for the cc_config.xml.

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours

Out of interest how old is your 295?

---

Yes skildude i´ve turned windows aero off and in visual effects, I set it to have better performance, hoping for less video card memory use.
And yes, fitting the card in my other machines, it does the same thing.
No, by using more than 500mb (503 to 514, depending the instaled driver) I mean I use EVGA Precision to oc and monitore the card, and it reports that amount of video card memory used. I know there is plenty of memory available (over 880mb each GPU) but still, using that much memory seems not to be right, and it seems to me its a memory related problem.

TY Gundolf for the tip.

Speedy, this GTX is like 8 months old, about the same as the other I have that is perfect.

---

Mine was new last August. It is good you had a good one to compare to. It took a long time for me to be convinced mine is actually defective, as I was not sure what to expect as I had never done GPU crunching before. I didn't even know the 295 was the hottest card out there at the time. With Gatekeepers experience, and yours, I am now sure some of these cards were defective from the get go, but because they play games just fine, Nvidia seems to be paying attention to the other items on their plate.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

I'd say that the card is shot. If you get repeated results on different machines. This is pretty obvious. It's time to return the card

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I'd say that the card is shot. If you get repeated results on different machines. This is pretty obvious. It's time to return the card

I did it. But as scimanstev says, if the card works in games (wich it does), they won´t replace im. The card was almost 45 days in RMA service, and they refuse to change it.
I need to find a game, 3d mark or somethng that show the card is failed, I guess.
But I really will keep trying to make it work like one or two weeks before that...

---

there was a gpu stress test that someone has linked a few months ago. perhaps they'll point it out again.

have you looked at GPU-z and checked the GPU temperatures under stress? My fist ATI 5850 ran hot and failed after a couple weeks. My new card runs much cooler even under stress

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

Yep, temperatures are ok, like 70°C under full load, and power is ok too (seventeam 850w power source).
Heh... can´t make that ignore cuda device command work... omg I really sux at software skills...

---

Heh... can´t make that ignore cuda device command work...

The cc_config.xml file must go into the BOINC data directory (see Messages tab after restart of client).

Gruß,
Gundolf

there was a gpu stress test that someone has linked a few months ago. perhaps they'll point it out again.

have you looked at GPU-z and checked the GPU temperatures under stress? My fist ATI 5850 ran hot and failed after a couple weeks. My new card runs much cooler even under stress

The software is called Furmark. Here is a link.

http://www.ozone3d.net/benchmarks/fur/

My card did fine with this intensive test, but won't crunch properly with optimized apps. On regular apps, it works fine.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website