Hi friends,

I'm sitting here and watch my (single .. <sigh!>) NVIDIA RTX 2080super crunching data.

Interstingly I see that my machine is capable of (alongside running Rosetta@home WUs on the CPU)

- cruncing a BOINC project (Milkyway, SETI, Einstein)
- running folding@home on the GPU and
- do some GPU-accelerated video transcoding

concurrently (!).

The only project that's capable to wind up my GPU fans to audible level all alone is SETI@home. I wonder if it's possible to stuff more than one BOINC task into that GPU ?

Any hint ?

Seneca
0=0

---

I wonder if it's possible to stuff more than one BOINC task into that GPU ?

Yes, that's possible.
Either via an app_config.xml file (documentation: https://boinc.berkeley.edu/wiki/Client_configuration#Project-level_configuration), or some projects (such as Einstein) give the option via their project preferences (https://einsteinathome.org/account/prefs/project)

But you run FAH on the GPU, so if you want to run one more of their tasks on the GPU, you'll have to ask them if that's possible. FAH doesn't run under BOINC.

---

But you run FAH on the GPU, so if you want to run one more of their tasks on the GPU, you'll have to ask them if that's possible. FAH doesn't run under BOINC.

I'm aware of that ... (it nags me to be precise ... why don't they use BOINC ?).

Thanks for the tip, I'll investigate that.

Greetings

Seneca
0=0

By the way - any chance to run projects parallel on the GPU (not round-robin) ?

Seneca
0=0

(it nags me to be precise ... why don't they use BOINC ?).

That's something you'll have to ask them. Perhaps in the near future that's possible, seeing Greg Bowman's remark on the question if FAH shouldn't run under BOINC.

By the way - any chance to run projects parallel on the GPU (not round-robin) ?

You mean two tasks from different projects at the same time? I don't think that's possible on a single GPU, you'd always need a minimum of two GPUs then - and then run a task per GPU, or two tasks of the same project per GPU.

By the way - any chance to run projects parallel on the GPU (not round-robin) ?

Never mind - it intermixes the tasks if there's enoughGPU share for some task ... I'm satisfied ;-)

Running multiple WUs under 1 GPU doesn't always go without problems. For instance the project Moo!Wrapper (or Milkyway, one of the two, forgot which) uses less than 80% of my GPU's resources, regardless of how many WUs I run per GPU. However projects like Einstein allow you to fine tune the WUs (to run heavier units per GPU).