Hi everyone,

Hopefully someone can help or point me towards a thread that might already cover the problemo. Not sure if it is a CUDA-specific issue, CUDA workunits are at the core of it, but it's more a question regarding timely reporting of failed workunits.

In brief: I had 50 or so CUDA workunits lined up to work through earlier today, when I (admittedly absent-mindedly) updated my graphics driver. This caused BOINC to fail to detect my CUDA device and then deleted all my CUDA workunits. Go me, definitely learnt a lesson there.

Now I have these 50-odd CUDA workunits that are still "In Progress" on my results page for seti@home, all due on the 13th of November, but they are no longer on my computer. I know it isn't that big a deal as come the deadline a further set of duplicates will be sent out to make up for my failure to report in.

I was just wondering if there is any way for me to signal the project that these units are no longer being crunched by me so they can be sent out to others sooner, or is it just a matter of waiting till the deadline and it'll happen automatically?

I don't really care about the credits or anything, more that someone else (or even I) could be processing these tasks now rather then wait 2 weeks till the deadline.

Edit: I'm running BOINC 6.10.17, WinXP SP3 and my graphics card is a NVIDIA GeForce 8800 GTS (319MB) driver: 19539.

Cheers,

Taylor "Flake" Foster

In boinc manager, set the project to "No new tasks" and let it finish anything that it has at the moment.

Once these are done, detach from the project and then attach to the project again.

Once you attach again the server will reset all the workunits that you used to have and send them out to others.

---

Great, thanks for the help!