The link in the Anandtech forum points to the ExtremeOverclocking leaderboards for Folding@Home, which is a Stanford project to do distributed protein folding. They're currently placing COVID-19 related projects at max priority.
My understanding is that Amazon is devoting a chunk of spare AWS capacity to folding for the project. In the last week or so, they've done about 300k work units (essentially they've run that many simulations all the way through).
Let's be clear here: this is about protein folding.
I almost panicked for a moment thinking they were folding in some sort of business sense, before realizing it would take far more than what's currently going on for AWS to fold.
As usual ... Wikipedia [1] has a better explanation than the actual site [2], sigh.
"Folding@home (FAH or F@h) is a distributed computing project for performing molecular dynamics simulations of protein dynamics. Its initial focus was on protein folding but has shifted to more biomedical problems, such as Alzheimer's disease, cancer, COVID-19, and Ebola."
The teams are quite fascinating. A diverse bunch of national teams, and community teams from tech websites to the skeptics community to Bronys to Free Republic, the Republican mega-forum?
So I’m curious, we have a tendency on HN to assume every problem has a technical solution; from autism to inmate rehabilitation, from food insecurity to poverty, solutions are just one good SaaS idea away.
Is covid@home really going to solve this by donating cpu cycles before traditional medical research creates a vaccine? Is turning coal into code (ala Bitcoin) that productive in this case? Because I’ve run the app but feel like I’m in some kind of techbro bubble to believe I’m making even an iota of difference
Based on the "Results" page on the F@H site [0], they have managed to produce some reasonably useful results over the years. At first glance, it definitely seems like what they're doing could be helpful for COVID-19, as they're currently in the exaflop range of available computing power and are using it to identify druggable targets on some of the virus' proteins.
(I'm paraphrasing people here and I may have mangled what they're saying, sorry)
This thread has some useful commments, one from someone who's actually done F@H research, another from someone who mentions the length of time it takes to get a medication trial through the system (without covid-19 easements) and how few trials actually make it.
There's a bit of discussion around the cost in electricity, and whether it's better to just send that money somewhere.
This post is from someone who is a researcher who says that F@H is a form of science, and so it should continue, but that it's probably not as directly applicably useful as F@H says.
COVID-19 High Performance Computing Consortium
Members: a lot of universities, federal agencies, NASA, US DoE, IBM, Amazon Web Services, AMD, Google Cloud, Hewlett Packard Enterprise, Microsoft, NVIDIA,
compute muscle -> Over 402 petaflops, 105,334 nodes, 3,539,044 CPU cores, 41,286 GPUs, and counting.
24 comments
[ 2.8 ms ] story [ 66.1 ms ] threadI'm still not sure what this is. Who is ec2spot??
My guess is that maybe Amazon installed the Folding client on some part of ec2 capacity in an idle pool waiting to be launched?
I'm not sure. That would use quite a lot of power... and setting up networking and getting approval to do this would be a ton of work.
My understanding is that Amazon is devoting a chunk of spare AWS capacity to folding for the project. In the last week or so, they've done about 300k work units (essentially they've run that many simulations all the way through).
I almost panicked for a moment thinking they were folding in some sort of business sense, before realizing it would take far more than what's currently going on for AWS to fold.
"Folding@home (FAH or F@h) is a distributed computing project for performing molecular dynamics simulations of protein dynamics. Its initial focus was on protein folding but has shifted to more biomedical problems, such as Alzheimer's disease, cancer, COVID-19, and Ebola."
1: https://en.wikipedia.org/wiki/Folding@home
2: https://foldingathome.org/
https://folding.extremeoverclocking.com/team_list.php
HN should start one.
I suggest changing the title to Folding@home, the more common name for this type of project.
Is covid@home really going to solve this by donating cpu cycles before traditional medical research creates a vaccine? Is turning coal into code (ala Bitcoin) that productive in this case? Because I’ve run the app but feel like I’m in some kind of techbro bubble to believe I’m making even an iota of difference
[0] https://foldingathome.org/papers-results/
This thread has some useful commments, one from someone who's actually done F@H research, another from someone who mentions the length of time it takes to get a medication trial through the system (without covid-19 easements) and how few trials actually make it.
There's a bit of discussion around the cost in electricity, and whether it's better to just send that money somewhere.
https://news.ycombinator.com/item?id=22491910
This post links to a blog that talks about F@H that you may be interested in.
https://news.ycombinator.com/item?id=22662026
This post is from someone who is a researcher who says that F@H is a form of science, and so it should continue, but that it's probably not as directly applicably useful as F@H says.
https://news.ycombinator.com/item?id=22576088
There is also https://covid19-hpc-consortium.org/
compute muscle -> Over 402 petaflops, 105,334 nodes, 3,539,044 CPU cores, 41,286 GPUs, and counting.To join - go to https://www.xsede.org/covid19-hpc-consortium
Just curious - are Apple/FB doing anything about it?
EDIT: Apple is helping in a different way (https://www.theverge.com/2020/4/5/21209270/apple-face-shield...)