I only seem to be downloading 1 WU at a time, but as I have a GTX 270, I can get through it in less than 10 hours. Unfortunately, no more WUs download. Hitting "update" only gets "communication deferred 30 secs" and nothing ever happens. So I have to detach and re-attach to the project to get another. Is GPUGrid not entirely compatible with BOINC 7.0.18 x64 (which I'm running)? Or is there something else going on?

EDIT: My global settings are set to maintain 5 days' worth of work, and I have plenty of hard drive space allotted.

GTX 570.
New apps being released means current tasks are running down. New apps don't turn up too often. Normally the maximum number of tasks you get is 2 anyway. It's much better for the project to return tasks quickly, rather than have them sit on your system for days, hence the maximum is 2tasks.
Ideally the number of tasks you have in queue should be balanced against your GPU's performance and the Long/normal task you crunch for, but such things cant be done easily.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Yes, GTX 570. Sorry.

I was hoping to use the GPU exclusively for GPUGRID, but I can't if it won't automatically download new WUs when it's finished. I have the "report results immediately" flag set in cc_config, and it does submit the work shortly after it is complete. Isn't there a way to get more than 1 or 2 at a time? Manual WU download? Anything?

Also, I've only been getting the Long Run WU, and they seem to only use ~70% of my GPU (I have a little script that automatically adjusts the process priority to "normal" from the default "below normal", but 70% is the max it ever gets to). Is there a setting somewhere that I'm not aware of that would enable closer to 100% use? Or maybe run 2 WU in parallel? I have yet to come across a functional app_info for GPUGRID that would allow that.

I see you are using the Boinc Alpha version BOINC version 7.0.18.
For that the settings are for a minimum work amount and a maximum work amount. If you want to make sure there are tasks on the system increase the minimum work amount to say 0.5 days and the maximum to 1.0 days. You could also use 0.25 and 0.5 which would give you a greater chance of getting the credit bonus for completing tasks within 24h.

"Normal" makes no difference over "below normal". Freeing a CPU core and using SWAN_SYNC might improve performance slightly. Process Lasso might also help.

GPU utilization varies between different tasks. On average most tasks are higher than 70%. Nate's were ~99% last time I looked.

Have you renamed the system &/or repeatedly reattached?
http://www.gpugrid.net/results.php?hostid=119583
http://www.gpugrid.net/hosts_user.php?userid=86296

You might want to merge systems by name.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Where do I set the minimum and maximum work amounts? My computing preferences for BOINC say that I'm always connected to the internet, and I should maintain enough work for 2 additional days.

Yes, like I said, I have had to reattach to get new WUs. I went ahead and "merged" the computers via your link (thanks).

I still only seem to be getting 1 WU at a time, then nothing until I reattach.

2 errors, 3 timeouts and 2 in progress, that were presumably deleted by detaching.

http://www.gpugrid.net/results.php?hostid=119722&offset=0&show_names=1&state=0&appid=

http://www.gpugrid.net/host_app_versions.php?hostid=119722

Then there is a new app which might change the device rating further.

Leave the settings alone for a few days to allow them to balance out. Detaching and reattaching is counterproductive.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help