Tasks runs fine on 560Ti and 260.

Noleias run on 1GB GPUs but not on 768MB cards. The Santi and Nathan WUs run great with 768MB. Unfortunately I can't get any for the GPU that needs them.

I have just had a couple die on my 580.

This one crashed completely - http://www.gpugrid.net/workunit.php?wuid=4854487
This one crashed, too, but there is no indication to the server that it died - http://www.gpugrid.net/workunit.php?wuid=4854479

Are these, perhaps, to problematic to be in the field?

I've crunched successfully a couple of them on my GTX 680s and GTX 670s, and I'm crunching one on my GTX 480 for 10 hours (less than 2 hours left).

GTX680 and GTX460 (1gb) are finishing the NOELIA WU without errors.

I have just had a couple die on my 580.

Are these, perhaps, to problematic to be in the field?

I have 4 NOELIA_FXA running on my dual GTX590 box now and have had no problems running the ones I've been sent (4 completed and validated so far).

http://www.gpugrid.net/results.php?hostid=159990&offset=0&show_names=1&state=0&appid=

Win7x64 and 331.40 drivers.

Operator

---

Interesting, your CPU time is about 1/4th of the run time, mine are almost the same (run time a few seconds more than CPU time).

---

Greetings from TJ

Im not sure, i rebooted today one of the machines with this wu type duo it run over 24hours and was on 98% after reboot it disappear on boinc and gpugrid -_- next time i wait the last 2%..
You can see the timegap here http://www.gpugrid.net/results.php?hostid=117426

---

DSKAG Austria Research Team: http://www.research.dskag.at

Interesting, your CPU time is about 1/4th of the run time, mine are almost the same (run time a few seconds more than CPU time).

The Fermi based cards (GTX4xx and GTX5xx) have the CPU time much lower than the Kepler based cards.

See http://www.gpugrid.net/forum_thread.php?id=3396&nowrap=true#30750

Interesting, your CPU time is about 1/4th of the run time, mine are almost the same (run time a few seconds more than CPU time).

The Fermi based cards (GTX4xx and GTX5xx) have the CPU time much lower than the Kepler based cards.

See http://www.gpugrid.net/forum_thread.php?id=3396&nowrap=true#30750

That's because Kepler cards take up a CPU core by design like using swan_sync=0 for earlier cards to achieve the same thing.

Thanks Jeremy and Betting Slip, now I know why.

---

Greetings from TJ

Speedy wrote:

Can somebody please explain to me what is meant by these 4000 tasks only having 2 steps?

Stefan wrote:

...
Technically for our users I think it's of little significance how many steps the simulations consist of.

The significance of the 2 steps is that when we crunch this batch of workunits, they will wear out from the queue rapidly, because every workunit produce only another one.

Getting some more Noelia FXA wu's since the server was topped up. I thought this project had finished?

I've received two Noelia FXArep as well.

We realized we needed one last bunch of WUs to finish the project. We've already started writing, so should be able to post final results sometime soon, hopefully :)
thanks!!

I've received two Noelia FXArep as well.

I got already 8 of them and they run smooth on my systems with a massive GPU load of 95%.

---

Greetings from TJ

We realized we needed one last bunch of WUs to finish the project. We've already started writing, so should be able to post final results sometime soon, hopefully :)
thanks!!

You send them Noelia, we'll crunch them!!!
I have always been, and continue to be, impressed with the GPUGrid team and the work you do. It feels good to have the results of my hobby actually be useful in helping you reach real goals.

---

Thanks - Steve

+1

---

BOINC <<--- credit whores, pedants, alien hunters

Just checked both my rigs and all 6 GPUs are running these.

---