Last week I did run a few WU's on my nVideaGTX285.
The last one resulted in a blue screen for three times after rebooting. All other things were running fine afterwards the GPUGRID was cancelled and the pc run for 60 hours now without problem.

When however I ask new work it is saing the beta version is not available fo my pc. That is normal (no Tesla). However other work is available according the server status page. Have more people this issue?

Thanks.

---

Greetings from TJ

I have the same issue but my other rig with the same spec and config can download new wu.

65 GPUGRID 11/21/2010 02:53:20 PM Message from server: No work sent
66 GPUGRID 11/21/2010 02:54:56 PM Sending scheduler request: To fetch work.
67 GPUGRID 11/21/2010 02:54:56 PM Requesting new tasks for GPU
68 GPUGRID 11/21/2010 02:54:58 PM Scheduler request completed: got 0 new tasks
69 GPUGRID 11/21/2010 02:54:58 PM Message from server: No work sent
70 GPUGRID 11/21/2010 02:57:33 PM Sending scheduler request: To fetch work.
71 GPUGRID 11/21/2010 02:57:33 PM Requesting new tasks for GPU
72 GPUGRID 11/21/2010 02:57:35 PM Scheduler request completed: got 0 new tasks
73 GPUGRID 11/21/2010 02:57:35 PM Message from server: No work sent

---

Can you make your machines visible?

gdf

No they will stay hidden.
The information it gives is also limited.
The software and the hardware isn't changed in the last 6 months exceot windows vista upgates.
All projects run smootly, your did too. But now I don't get new WU's.
I get thes messae:
02/12/2010 12:44:03 GPUGRID Message from server: No work sent
02/12/2010 12:44:03 GPUGRID Message from server: ACEMD beta version is not available for your type of computer.

That's correct but I got "the normal" work previously.

Intl i7, 64Vista Ultimate, GTX 285, driver 19107.
And with that Milyway run flawlessly 24/7.

---

Greetings from TJ

The 19107 driver looks too old.
Update the driver to a newer driver between 195.xx and 19745 (inclusive).
You should then download the 6.12 GPUGrid app and run tasks for it.

Thanks Skgiven for a possible solution.

However two weeks ago it ran with this driver. I will update in due course.
The app i have also never downloaded before, everything went automatically.

---

Greetings from TJ

Thanks Skgiven for a possible solution.

However two weeks ago it ran with this driver. I will update in due course.
The app i have also never downloaded before, everything went automatically.

What skgiven means is that once you have installed the new driver, the new application will automatically follow it, just as the previous application was downloaded automatically when you had the older driver.

The new application, which requires the use of the newer driver, was deployed on 3 November - slightly longer ago that the two weeks you remember - but I can't find a clear and explicit announcement this long after the event.

I think some tasks employed new subroutines subsequent to that app release, but definitive anouncements might have helped in this case (overall there are so many changes that it would impact on the science just to anounce each of them).

Well I have upgrated to the latest nVidea drivers but still no new work.
MilkyWay is now running with the graphics card smoothly.

Perhaps if you suspend MW tasks you will start to pick up GPUGrid tasks.
Exactly what driver did you install?
Did you restart after the update?
How much RAM do you have?
How much free disk space do you have?
How many tasks do you have in queue and what is your queue time, CPU usage?

There were no takss, just waiting 30 mins for yours.
NVIDIA GPU 0: GeForce GTX 285 (driver version 26099, CUDA version 3020, compute capability 1.3, 998MB, 708 GFLOPS peak).
Yes I restarted.
12Gb Ram.
374Gb free disk space.
Nil in queue, when waiting yours, now 3 Einstein running. CPU usage is currently 44%.
I also updated BOINC to 6.10.58 and did a second restart after that.
Vista 64 bit ultimate.

Any ideas?
Thanks.

---

Greetings from TJ

Hello all,

I just got a new WU from GPUGRID, so I am in medicine again.
It's a KASHIF_HIVPR and I have read it errors out after a few hours. Let's see what it does on a GTX285.

---

Greetings from TJ

I have read it errors out after a few hours.

Where?
These tasks work fine.

I have read it errors out after a few hours.

Where?
These tasks work fine.

Hi skgiven,

I read it in the threat "lots of failures" about in the middle.
You have also answerd a few.
And in this threat I have mentioned to wich driver I have updated.

---

Greetings from TJ

Your credit and RAC seem to have risen, so I guess you are at least finishing tasks.

I suggested you use a newer driver, as the one you had was not new enough to get any tasks. I suggested up to 197.45 to run the 6.12app.
If you upgraded to a newer driver than that you will be running the 6.13app. This is OK, but these more recent drivers might cause your GPU to run at reduced speeds. Again, please confirm what driver you installed.

PS. It is unlikely that advice aimed at another user with a different card in another thread is going to be correct for you.

skgiven,

4 Messages bewlow you can see what is installed.
You asked yesterday and I did replay.

Indead 2 WU's are ready with cedit and number 3 is running.
So all works fine again.

---

Greetings from TJ

Two CA's suggested you install 197.45, and explained why, but you installed 260.99!

Your RAC is still rising, so it seems to be working. For your card (and I speak generically here) there are less problems with that driver on Vista than on XP, so you might get away with keeping the driver.
One thing I would suggest is that you right click on the desktop, click NVidia control panel, manage 3D settings, global settings, Power Management Mode, and select so that the GPU is set to run at maximum performance (prefer maximum performance). If its not selected some cards go into an energy saving mode and drop their GPU & shader frequencies to the lowest possible, causing tasks to take around three times as long to run. This happens almost randomly, so it could be fine for a day or so, then happen every couple of hours. This is the main reason for not using the 260.99 driver. The second reason for your card, is that by using it you will run the 6.13app which is slightly slower for your card. It's faster/required for 400series Fermi's.

PS. I find trawling through two threads to find system specs not as easy as clicking on a users ID and having these things listed, along with success/failures, GPU speed and runtime. These tell me a lot more than everything you posted in the threads.

Good luck,

"No work sent"

We have all seen this but I don't recall seeing a definative answer from GPUGrid as to what the *rules* for this behavior is. Can someone who knows what the rule really is please let us know?
( no guessing allowed :p )

I experience this whenever a WU crashes ... GPUGrid shuts me off from getting new WUs for some indeterminate amount of time. I understand that you need to shut down machines that do nothing but spew errors but my 480GTX is pretty very stable.

This happens even if it is a bad WU like http://www.gpugrid.net/workunit.php?wuid=2193663

I have observed that if you have another WU alreeady in your queue you have to finish that one sucessfully before you get another.

Keep up the great work on the science ... the rest of us will keep crunching!!!

Thanks in advance for your time,

---

Thanks - Steve