As far as I remember we discovered and reported this while CUDA 4.0 was in alpha test to developers, I think about 2.5y ago. We tested and reported with every new CUDA version that has come out since.
This was reported in the standard bug reporting & tracking system available to registered developers at NVidia, actually by Oliver who keeps track of that issue on out side. I don't remember if one of our contacts at NVidia was pointed to that issue in particular.
I have access to nVidia NDA Partner's Portal and can submit issues thru that system. I could submit this problem if you wish, maybe it would get more attention. Let me know...
I'm a bit nervous about the length of these tasks combined with my using my GPU for other purposes. Occasionally when I either forget to suspend the GPU while gaming, or when BOINC decides to unsuspend the GPU while I'm gaming, instead of just slowing to a crawl BRP4 tasks end up crashing. Losing a partially complete 20m task wasn't a big deal; losing several hours of work is a more serious concern.
That said, I'm not sure if there's anything you can do about the problem. IMO modernzing the boinc client's exclusive app feature would be the proper fix. All or nothing behavior was good enough when most systems were single core; now with consumer CPUs being able to run 8 threads in parallel but many games still only using 1 or 2 heavy threads it's no longer adequate.
is it for nvidia only? or ati can participate too?
I got 1 on my NVidia GT240 estimated runtime > 13 hours and
1 on my ATI 7770 with an estimated runtime of almost 5 hours.
No overclocking and single GPU-task selected.
Thanks Bernd for the reply.
)
Thanks Bernd for the reply. Let's hope they fix it someday, so the apps can run more effective. For how long is this problem reported to nVidia ?
-----
As far as I remember we
)
As far as I remember we discovered and reported this while CUDA 4.0 was in alpha test to developers, I think about 2.5y ago. We tested and reported with every new CUDA version that has come out since.
BM
BM
Uh, that's way too long.
)
Uh, that's way too long. nVidia is not an easy company to deal with...
Where/to whom have you reported to issue?
-----
This was reported in the
)
This was reported in the standard bug reporting & tracking system available to registered developers at NVidia, actually by Oliver who keeps track of that issue on out side. I don't remember if one of our contacts at NVidia was pointed to that issue in particular.
BM
BM
I have access to nVidia NDA
)
I have access to nVidia NDA Partner's Portal and can submit issues thru that system. I could submit this problem if you wish, maybe it would get more attention. Let me know...
-----
I'm a bit nervous about the
)
I'm a bit nervous about the length of these tasks combined with my using my GPU for other purposes. Occasionally when I either forget to suspend the GPU while gaming, or when BOINC decides to unsuspend the GPU while I'm gaming, instead of just slowing to a crawl BRP4 tasks end up crashing. Losing a partially complete 20m task wasn't a big deal; losing several hours of work is a more serious concern.
That said, I'm not sure if there's anything you can do about the problem. IMO modernzing the boinc client's exclusive app feature would be the proper fix. All or nothing behavior was good enough when most systems were single core; now with consumer CPUs being able to run 8 threads in parallel but many games still only using 1 or 2 heavy threads it's no longer adequate.
The first ~5000 tasks of BRP5
)
The first ~5000 tasks of BRP5 are being sent out on Einstein.
BM
BM
RE: The first ~5000 tasks
)
Great news :)
is it for nvidia only? or ati
)
is it for nvidia only? or ati can participate too?
RE: is it for nvidia only?
)
I got 1 on my NVidia GT240 estimated runtime > 13 hours and
1 on my ATI 7770 with an estimated runtime of almost 5 hours.
No overclocking and single GPU-task selected.