That sucks. Please tell me when the issues have been resolved so I can re-enable beta.
18224 will not make the timeout (Radeon 780M)
Moderators: Site Moderators, FAHC Science Team
Re: 18224 will not make the timeout (Radeon 780M)
Hey arisu, sorry an 18224 made it to your GPU. We've got additional restrictions in place now to prevent this. I believe that 12705 has been adjusted though I'm not the project manager for that project.
To your previous question:
To your previous question:
WUs are sent out according to individual project priorities (e.g. 18824's could be different than 12705), but are some combination of the Run, Clone, and Gen #. E.g. one could prioritize completing all GEN0 runs in CLONE 0, then all GEN1 runs in CLONE0, etc. Once the timeout expires, the WU will go back into the "available to send out" pile and the next time it reaches the top priority WU it will be sent out. If a project is in high demand, it's likely the WU will go back out immediately. If there's been a sudden drop off in demand and the WU is low priority in the ranking system it could take a good deal of time.Actually, how long after the timeout expires does a duplicate WU get sent out? Is it immediate during the next credit check or is there a delay? If there is a delay then I'll just let the timeout expire naturally, because in both cases it will finish within half an hour of the timeout expiring.
-
- Posts: 1661
- Joined: Sun Dec 16, 2007 6:22 pm
- Hardware configuration: 9950x, 7950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, RX 550 640SP - Location: London
- Contact:
Re: 18224 will not make the timeout (Radeon 780M)
Re: 18224 will not make the timeout (Radeon 780M)
Awesome thanks.
I finished writing my blacklist patch, and it's working. This should help me avoid projects that would result in sending out duplicate work, so science doesn't have to suffer more because of OpenCL.
(EDIT: copy of patch removed after I was alerted that it may enable cheaters)
It creates a new option "project-blacklist" with a space-delimited list of projects. Projects are not dumped, just rejected as if the connection failed during download (which happens to me a LOT anyway).
I finished writing my blacklist patch, and it's working. This should help me avoid projects that would result in sending out duplicate work, so science doesn't have to suffer more because of OpenCL.

(EDIT: copy of patch removed after I was alerted that it may enable cheaters)
It creates a new option "project-blacklist" with a space-delimited list of projects. Projects are not dumped, just rejected as if the connection failed during download (which happens to me a LOT anyway).