Distributed scientific computing, do your part (winter's the perfect time)

Liberty · November 22, 2018

Your yearly reminder that the cold season is the perfect time to run distributed computing projects on your CPUs/GPUs/consoles. Extra heat produced just heats your house... My fave is Rosetta@home, a biomedical computational protein design project:

https://boinc.bakerlab.org/

Feature piece about David Baker, the head of Baker lab where Rosetta@Home is based:

http://discovermagazine.com/2018/nov/all-in-the-fold

Another popular project in health research of Folding at Home. Because of what it does, it's particularly efficient to run on GPUs and consoles:

https://foldingathome.org/start-folding/

(this one has a version that can run in-browser, so you don't even need to install the app to help: http://nacl.foldingathome.org/ )

But there are many projects, some in physics, math, cryptography, etc:

https://en.wikipedia.org/wiki/List_of_distributed_computing_projects

Einstein@home's a cool one that looks for gravity waves:

https://en.wikipedia.org/wiki/Einstein@Home

On August 12, 2010, the first discovery by Einstein@Home of a previously undetected radio pulsar J2007+2722, found in data from the Arecibo Observatory, was published in Science.[4][5] The project had discovered 55 radio pulsars as of November 2016.[6][7]

As of November 2016, Einstein@Home has discovered 18 previously unknown gamma-ray pulsars[8] in data from the Large Area Telescope on board the Fermi Gamma-ray Space Telescope.

Hielko · November 22, 2018

I doubt that running this at home is an efficient use of energy.

Liberty · November 22, 2018

I doubt that running this at home is an efficient use of energy.

How so?

I live in Canada. I have to heat my house. All electricity (hydro, in my case) used by this is then converted to waste heat, which helps heat my house.

This is just a smaller version of baking in the winter (you get food and heat -- cogeneration?) reducing how much the furnace has to run...

clutch · November 22, 2018

I think this idea is pretty cool. Although it should give me some (even minimal) incentive for me to leave my computer on to run the software. If not money, some gamification / reward of some sort would be nice. E.g., if you are in the top 10 contributors of all time, you will be acknowledged in some future publication.

Also, I guess security could be an issue. Hopefully these developers spent enough time making sure the software can't be hacked to steal other data from your computer.

Liberty · November 22, 2018

I think this idea is pretty cool. Although it should give me some (even minimal) incentive for me to leave my computer on to run the software. If not money, some gamification / reward of some sort would be nice. E.g., if you are in the top 10 contributors of all time, you will be acknowledged in some future publication.

Also, I guess security could be an issue. Hopefully these developers spent enough time making sure the software can't be hacked to steal other data from your computer.

Both of those concerns are addressed.

There are teams, points (more computations = more points), there are APIs, so dozens of sites give you different visualizations of your stats (there's a better ecosystem than for financial data, almost). You get daily average number of points, if you run it on more than one computer you can use the same account so it's cumulative, etc.

And it's been around for a long time, it's open-source, and it's based at Berkeley university. They take security pretty seriously and are totally legit.

"Folding@Home is supported by the NIH and NSF, and already has over 200,000 active users. It has been published in over 100 papers"

Rosetta also has published many papers, including in Science and Nature.

https://www.bakerlab.org/index.php/publications/

Jurgis · November 22, 2018

I doubt that running this at home is an efficient use of energy.

How so?

I live in Canada. I have to heat my house. All electricity (hydro, in my case) used by this is then converted to waste heat, which helps heat my house.

This is just a smaller version of baking in the winter (you get food and heat -- cogeneration?) reducing how much the furnace has to run...

I think the possible critique is that it's not efficient use of your computer/money. Assuming running programs on computer increases risk of failure (vs computer sleeping - I think this is true), it is likely much more economic to donate $X00 to run this on blades in data center rather than risk a failure of your cooling system/motherboard/CPU/GPU/HDD that would cost more in money and inconvenience to replace. Especially on laptops where piecemeal replacement is hard.

The goals are worthwhile so I am not trying to dissuade people from doing it. I've done SETI@home 15 years ago or so.

Whether you run these or not: backup often and well. 8)

SHDL · November 23, 2018

A previous employer of mine had this type of thing set up internally so that you could submit batch jobs to a scheduler and then it would automatically allocate them to idle workstations within the organization. It was a pretty neat setup that allowed them to run massive Monte Carlo simulations and the like on a limited budget in those pre-AWS days.

randomep · November 23, 2018

I doubt that running this at home is an efficient use of energy.

How so?https://drive.google.com/drive/my-drive

I live in Canada. I have to heat my house. All electricity (hydro, in my case) used by this is then converted to waste heat, which helps heat my house.

This is just a smaller version of baking in the winter (you get food and heat -- cogeneration?) reducing how much the furnace has to run...

But conversely if I live in Arizona, the heat would cause me to run up the AC, so americans should offload their computing to canadians!

Liberty · December 3, 2018

This is cool:

https://deepmind.com/blog/alphafold/

LC · December 3, 2018

This is cool:

https://deepmind.com/blog/alphafold/

Very interesting, thank you

Cigarbutt · December 3, 2018

Yes, interesting work with potentially short term applications.

For instance, in the world of biosimilar insulins that result from DNA recombinant technology, a very minor change in sequencing may have a huge effect on the efficacity and side effect profile of the drug because of the way the molecule folds on itself in a 3-dimensional kind of way.

Liberty · December 10, 2018

This is cool:

https://deepmind.com/blog/alphafold/

One person's commentary on what happened at CASP with AlphaFold:

https://moalquraishi.wordpress.com/2018/12/09/alphafold-casp13-what-just-happened/

Liberty · December 9, 2019

Good time to bump this up. I currently have 13 CPU cores running Rosetta@home 24/7.

Here's a recent presentation by Dr. David Baker about computational protein design (he's the head of Baker Lab at U. Washington, where Rosetta software is being developed):