GPUGrid Work Availability
- Thread starter Nick Name
- Start date
There's now some experimental Python work available.
www.gpugrid.net
Experimental Python tasks (beta) - task description
The certificate has been updated. All I needed was to do was restart BOINC.
There's some intermittent work, I think demand is exceeding supply. I've gotten a few Python tasks but haven't been able to successfully complete any. Most are failing on wingmen as well but some are completing. They barely load the GPU and are multi-threaded which is kind of annoying.
I meant to say earlier that a project reset was suggested if Python jobs were failing. I set to NNT so that I could do that, but as luck would have it I completed my first valid Python task today. 
Credit on these is just a fraction of the ACEMD tasks, and you can opt not to receive them if you wish. I'm not that concerned about the credit but even tasks that fail to validate tend to run for several hours.
Note that the current Python work runs mostly on the CPU even though it's called a GPU app.
Credit on these is just a fraction of the ACEMD tasks, and you can opt not to receive them if you wish. I'm not that concerned about the credit but even tasks that fail to validate tend to run for several hours.Note that the current Python work runs mostly on the CPU even though it's called a GPU app.
I haven't gotten any work here for a few weeks. A lull over the holidays isn't uncommon. If you're looking for another project in the meantime why not give Einstein a try.
Getting steady work now so get them while you can if you're interested. These tasks are BIG, taking nearly 12 hours on a 2080 Ti, so you probably shouldn't run this project on anything less than a 1070/80 GPU if you want to qualify for the 24 hour bonus.
These tasks are BIG, taking nearly 12 hours on a 2080 Ti, so you probably shouldn't run this project on anything less than a 1070/80 GPU if you want to qualify for the 24 hour bonus.
I haven't run one in over a month but they were all over the place in terms of run time, memory, disk space, CPU usage, and GPU usage. Credit was always 75K.
This task completed in 3 hours 15 minutes for 105k credits. The BOINC estimate was over 8 days but I knew not to take that too seriously. More concerning is the app spawned 32 instances of python.exe and maxed out the processor while GPU load was nil. This is clearly more of a CPU than GPU app. I had set to NNT to see how this task went. I'll run a few more and see how they go. My feeling is this has become a project best run on a dedicated system.
It looks like I'm going to have to upgrade the drive before I can get anymore work. It's an old SSD, frankly I'm surprised it still works. Its' only 50 gigs and GPUgrid is complaining it needs more space. I don't have a lot of free space left so I guess it's time to upgrade. It will be a little bit before that happens. Back to the original point of this thread: There is Python work available if you have the resources to run it.