Why no Core 22 work units
Moderators: Site Moderators, FAHC Science Team
-
- Posts: 80
- Joined: Tue Dec 19, 2017 12:19 pm
Why no Core 22 work units
Jeez, I want to do work. 1 3080 and 3 2080TI's sitting idle since only 600 core 22 jobs out there.
Re: Why no Core 22 work units
Weird stats indeed on https://apps.foldingathome.org/serverstats :
I guess someone is addressing the issues of the corresponding servers and people with fast GPUs are the first affected. It should get better soon.
Code: Select all
GRO_A8 117,161 0 117,161 6,444.00/hr
GRO_A7 230,893 0 230,893 6,420.00/hr
OPENMM_22 43 0 43 0.00/hr
Totals 348,055 0 348,055 34,308.00/hr
Re: Why no Core 22 work units
So do I.
I do undertand 3 facts:
1) FAH is about science; it's not a game. When there's no useful work that can be assigned to your GPU we don't reassign redundant work ("busy work") allowing you to appear to be producing valuable research results when you really aren't.
2) Good science requires deep though and good planning. Until a scientist completes those steps, you can't work on his project.
3) Various proteins differ from one another. Some work inefficiently on specific hardware. Small proteins are only able to use a small fraction of the shaders on your 3090 or 2080s and can just as easily be completed on a much smaller GPU (provided they're also idle). In that event would be inefficient if it is assigned to your GPUs. (You'd probably complain that you see your GPU is only N% busy where N is a small number). It's better to let your more powerful GPU wait until there's a big job for it to do. (See 1, above)
I do undertand 3 facts:
1) FAH is about science; it's not a game. When there's no useful work that can be assigned to your GPU we don't reassign redundant work ("busy work") allowing you to appear to be producing valuable research results when you really aren't.
2) Good science requires deep though and good planning. Until a scientist completes those steps, you can't work on his project.
3) Various proteins differ from one another. Some work inefficiently on specific hardware. Small proteins are only able to use a small fraction of the shaders on your 3090 or 2080s and can just as easily be completed on a much smaller GPU (provided they're also idle). In that event would be inefficient if it is assigned to your GPUs. (You'd probably complain that you see your GPU is only N% busy where N is a small number). It's better to let your more powerful GPU wait until there's a big job for it to do. (See 1, above)
Posting FAH's log:
How to provide enough info to get helpful support.
How to provide enough info to get helpful support.