Issues With Connecting To Servers

[muziqaz]

It is not weird for 5090. This GPU chokes with smaller projects and for it, everything but the most massive projects is small.
We can easily enable it for wider range of projects, but then you have people moaning that their 3000 dollar GPU is not getting enough PPD.
At the moment there is a shortage of GPU WUs overall

[arisu]

Damn, I didn't know that. Maybe I should rethink getting four 5080s (both 5080 and 5090 are species 10) and downgrade to something cheaper and more useful for science, since folding on something slower is better than not folding at all.

Is the GPU WU shortage a temporary thing or is this likely how it's going to be for a good while, with a surplus of GPUs? Just trying to rethink how best to optimize my contribution with the budget I have.

[muziqaz]

Shortages are like waves in the ocean They come and go. Not because of the more processing power available.

US government announcement of cut financing for viral research does not help things, either

[arisu]

I'll admit, this is a bit of a disillusionment. I was under the impression that FAH could always use more processing power, but if me spending money on good hardware will simply cause others, like OP here, to wait longer without actually speeding up the science at all, that's quite disappointing.

Maybe one day in a utopian future, enough force fields will be shared between GROMACS and OpenMM that there can be one core to rule them all, handling both GPU and CPU. Until then, it's not too late for me to send back my 5080s and buy some EPYCs.

[scorpuk]

Cheers for the heads up.

I’m only folding on the 5090 when I’m not gaming, so not a worry.

Is there a greater demand for cpu folding? I still have my 64 core opteron from the bigadv days gathering dust.

[arisu]

Cheers for the heads up.

I’m only folding on the 5090 when I’m not gaming, so not a worry.

Is there a greater demand for cpu folding? I still have my 64 core opteron from the bigadv days gathering dust.

Even though CPUs are much less efficient at folding than GPUs (by 10x to 100x), it does appear that there are more CPU jobs and more CPU jobs that are in the queue. The longer a WU waits in the queue, the higher the demand.

An Opteron might not be very fast. The 6380 gets ~25k PPD (https://folding.lar.systems/cpu_ppd/bra ... essor_6380). It's likely that your gaming computer's CPU will do more work, even if it's only folding while you aren't gaming. But every little bit helps of course!

[Dayle Diamond]

I've set my system for GPU folding only, and that's worked up until this recent shortage. Now I'm getting CPU work, which gets assigned but doesn't reach 1% until a GPU unit is also assigned, and they're assigned to me until I manually reject them. Not sure where to post this, as it might be an error, but it's not tied to one type of work unit.

[muziqaz]

I've set my system for GPU folding only, and that's worked up until this recent shortage. Now I'm getting CPU work, which gets assigned but doesn't reach 1% until a GPU unit is also assigned, and they're assigned to me until I manually reject them. Not sure where to post this, as it might be an error, but it's not tied to one type of work unit.

Post a screenshot of what you have set up with your clients