Folding Forum

The latest gromacs can use cpu+gpu on one WU.
But this is still experimental. It might not be beneficial in most cases.

jonault I find all your technical arguments reasonable and convincing.

The one idea that's interesting is that when a WU needs to be abandoned the partial results could be uploaded & handed off to someone else so they don't have to start over from scratch. I don't know how practical that is but it sounds promising. OTOH, computers that consistently fail to meet deadlines should probably not be encouraged to continue participating & this would run counter to that.

If even this alone can be implemented, actually all other suggestions of mine will become already fulfilled and no more necessary.

There would be no more need to break down large WUs into smaller ones for weaker computers because there will be no fear of "not finishing a WU on time". There will be no need to unify CPUs with GPUs to let them help each other because again there is no worry that "if each of them can't finish their task alone, all their contribution will just evaporate". And there would be no need to extend any deadlines because again there is no fear of expiration and data loss.

--------------------

You should remember that the ultimate goal of folding@home is not to give people something to to with their computers, it's to assist researchers in doing their research.

I don't find this in agreement with FAH's goal stated on their website. Actually, it is the other way around: their very goal is to use Excess power of computers for science; in other words, it is exactly to give people something to do with their computers and through that, help advance science and fight diseases.

Their website clearly reads:

We empower anyone with a computer and an internet connection to become a citizen scientist and join forces to fight global health threats by donating their unused computer power

On their home page, there is an emphasis on the words "ANYONE", "A COMPUTER" (i.e., any computer), and "UNUSED COMPUTER POWER". This is what makes F@H so attractive. The beauty of such distributed supercomputers is in the possibility of using the UNUSED capacity of average home computers running quite normally. This is like alchemy; it turns the already wasted time and computer power of an average user into science.

Otherwise, if it was the way you described it, F@H would be just another expensive supercomputer: Efficient but at the same time costing a lot of money, hardware, electricity, heat, etc.

Again: The only beauty of FAH to me is that it (supposedly) converts a lot of different sorts of "waste" to a lot of "pure gold", hence, alchemy. Otherwise, it would be no different from any other expensive, fancy supercomputer.

And if FAH is not what it claims on its website, then I think maybe they should revise their website.

--------------------

Old & slow computers that can't generate results in a useful time frame just aren't worth the effort to include because they aren't going to generate better results or faster results for the research, and the changes needed to include them are just going to slow everything else down.

I see. Still, WUs and projects can be categorized into smaller or larger ones, and weaker computers can be automatically assigned less demanding simulations. Therefore, their slower contributions would not bottleneck the whole system. And this means MORE DONORS.

Besides, if that all-or-nothing limitation is removed and even partial contributions are used by the system, then even slow computers can contribute without slowing down the whole system. That means a lot of new computers worldwide.

--------------------

On a side note: I see the website also states: "We need every ounce of computer power we can get!"

They need every ounce of computer power they can get but they lavishly discard tons of it. I hope the feature we agreed upon (not discarding expired WUs) gets implemented in near future.

vahid.rakhshan wrote: ↑Sat Feb 25, 2023 4:57 am On their home page, there is an emphasis on the words "ANYONE", "A COMPUTER" (i.e., any computer), and "UNUSED COMPUTER POWER". This is what makes F@H so attractive. The beauty of such distributed supercomputers is in the possibility of using the UNUSED capacity of average home computers running quite normally. This is like alchemy; it turns the already wasted time and computer power of an average user into science.

Yes, the website is very inspirational, but they've also said that they care very much about getting results as quickly as possible. Their research is an iterative process - try something out, see what happens, then get ideas from that about what to try next. So if you want to motivate them to make changes to include slower computers, I think you're going to have to come up with a convincing argument that doing so can generate end results faster so they can iterate more rapidly.

Personally, I think there's a balance between including more computers vs overall performance. Up to a point, adding more computers helps but at some point the slowest computers just become a drag on the whole system & you're better off without them. I don't know if the current cutoff point strikes exactly the right balance; I suspect that us folks on the outside doing the folding don't have enough information to answer that question.

I think perhaps we are not on the same page. Maybe my lengthy posts are tiresome to read. I will try to change fonts to capture your attention better. Hope I can make myself clear.

jonault wrote: ↑Sat Feb 25, 2023 5:06 pm ...but they've also said that they care very much about getting results as quickly as possible.

Faster results at what cost?

(Besides, I couldn't find any mention of caring very much about getting results as quickly as possible on their website. Care to share a link?)

And even if there was such a statement somewhere, it will clearly contradict their other claim that "anybody with any computer can become a citizen scientist by donating their UNUSED computer power".


----------------------------------------------

So if you want to motivate them to make changes to include slower computers, I think you're going to have to come up with a convincing argument that doing so can generate end results faster so they can iterate more rapidly.

I don't want to convince them at all that my suggestion will generate Faster results.

I am saying that (1) what they are currently doing is not what they are claiming on their website, and

(2) my suggestion of including the EXCESS computing power of so many more computers (even slower ones) will result in more EFFICIENT computations, i.e., greener, higher efficiency, more user-friendly, less demanding, less electricity waste, less generated heat, etc. by turning wasted time and resources into science.


Now, more efficacious computing may or may not accompany faster speeds. But that's not my point. My point is this:

Compare a Lamborghini with a highly efficient green car (for example, a car that runs on dumped and excess oil).
Of course, Lambo is so much faster but at the cost of a lot of expensive high-octane fuel and heat and noise and money.

The efficient car is not THAT fast, but it is still fast enough. It will take a bit longer to reach its destination, but it will consume only waste oil (instead of the high-octane fuel of Lambo) and will cost a lot less in every regard.

Now FAH pretends to be that "green car that runs on wasted oil as fuel". But in reality, it is a Lambo in sheep's clothing! I only want to remind them that this is not what they are advertising: a green supersystem of converting already wasted time and energy and computers into science.


----------------------------------------------



If they want to be very FAST at any cost, that's totally fine. Many researchers want or need to be Lambos. It is OK.This is why we have conventional supercomputers, in the first place.

But if FAH researchers want to be a Lambo, at least they can be honest with donors and disclose their ambition and say that they are not a green car, but an expensive and demanding Lambo. For example, they can say something like this:


Dear potential donors, the first and foremost priority of Folding at Home is speed. It does not intend to necessarily use the UNUSED power of your computers, but its main goal is to get results as fast as possible, at any cost, and this also includes running your computers at full speed 24/7 if needed. So if you want to join forces with us, be warned that we can and will be very very demanding to reach maximum powers at whatever cost. And if you have a slow or moderate computer, do not even think of becoming a citizen scientist!


----------------------------------------------

Yes, the website is very inspirational...

When this claim is not backed up by action, I wouldn't call it inspirational. I would call it misleading.


----------------------------------------------


ps. On a side note:

Imagine a green car running on waste oil that is optimized so well that it can go as fast as a Lambo or EVEN FASTER. Wouldn't that be awesome?

vahid.rakhshan wrote: ↑Sat Feb 25, 2023 10:03 pm (Besides, I couldn't find any mention of caring very much about getting results as quickly as possible on their website. Care to share a link?)

I would suggest reading the FAQ page on rules and policies - https://foldingathome.org/support/faq/rules-policies - especially the Best Practices section.

Also read the Points FAQ - https://foldingathome.org/support/faq/points. The section on bonus points is quite explicit:

Bonus Points

The prompt completion of Work Units (WUs) is very important for the science we’re doing. In order to study the proteins we’re interested in, we need be able get the results back quickly. A faster turnaround also means that we can launch projects that are larger and more difficult than ever before. So in 2010 we introduced the Quick Return Bonus (QRB), which gives extra points to users who rapidly and reliably complete WUs. The QRB has been fairly successful in aligning points with scientific value, and we will continue to use it.

There are other places on that site and here on this forum where this has been expressed collectively and by individuals in leadership roles in Folding@home. They are not hard to find if you actually look, and your dismissal of people who have been doing F@h for years telling you this does not go over well. There are other places where you misrepresent things related to F@h, you probably should start learning more about what you are criticizing first before commenting.

Now I get part of it, you found something you are enthusiastic about and think should be supported but it should be "better". But your views on "better" do not necessarily line up with the actual goals of the F@h projects. Ad in many cases what you propose here have been tried, found wanting, and F@h has moved on. Others have been not looked into, the researchers do not have the millions that might be required to implement many of the ideas proposed by you and others over the years.

Thanks for the links. They are helpful. And at the same time, they contradict what FAH advertises on its very homepage (as I guessed so yesterday).

Sorry but I find your tone somehow patronizing. Maybe it's my understanding, but I see lots of "you this, you that" and lots of projections and speculations and accusations. And your accusations are quite incorrect too.

Anyways, let's take something constructive from this discussion and move forward.

(1) The point is that that "better" I am enthusiastic about was advertised by FAH itself. Otherwise, if that "better" was not advertised by FAH in the first place, I wouldn't even know about it, let alone expect it or criticize not having it.

FAH can simply revise its home page and delete or change this: "We empower anyone with a computer and an internet connection to become a citizen scientist and join forces to fight global health threats by donating their unused computer power."

Voila! Everything solved! Again for the 100th time, I am not criticizing FAH for wanting to be fast or demanding. As I said repeatedly, that's up to them and also totally fine. I am just criticizing the part where they claim (rather falsely) that they need the UNUSED capacity of anybody's computer, while this is not really true.

(2) At least we agreed that one of the features I suggested can be very helpful, i.e., when I said unfinished WUs should not be discarded but instead be used by other users from where the previous donors left them off.

(3) Finally, it seems to me that given FAH's policy and your explanations, perhaps FAH doesn't even care about becoming more and more famous and known by casual donors. Perhaps, FAH prefers a limited group of competitive donors with very fast computers, rather than what I had in mind when I read the keywords "unused computer capacity" and "anyone" on its website.

vahid.rakhshan wrote: ↑Sun Feb 26, 2023 11:17 pm FAH can simply revise its home page and delete or change this: "We empower anyone with a computer and an internet connection to become a citizen scientist and join forces to fight global health threats by donating their unused computer power."

I would agree that that paragraph could use some revision.

Unused power is going to be interpreted by most as casual, not always running.
Most people have a slow laptop that is not well suited to fah.
Broadband internet is important.

At the time of Covid, we saw a huge influx of new folders.
At that time, deadlines were drastically shortened to keep everything running smoothly.
The influx of folders disappeared, but the deadlines stayed just as short.

I think a lot of new folders are jumping ship when they see that they can't meet the deadlines.

I don't know if you should accept unfinished calculations, increase deadlines or shorten calculation times, but for me, almost forcing folding all the time goes against the principle of the project.
And by the way, you're driving away a lot of folders and valuable computing power.

Thor wrote: ↑Mon Feb 27, 2023 9:33 pm I think a lot of new folders are jumping ship when they see that they can't meet the deadlines.

If people can't complete WUs in the required amount of time, they should move on to other distributed computing projects that are better suited to the hardware they have. This is a win-win - folding@home doesn't have to slow down and those other projects get resources they can use.

And if this results in so few people folding that projects can't get done in the desired amount of time, then the researchers will have to change their expectations. But until that happens there isn't a problem that needs to be solved - folding is getting done in the desired amount of time and that's what matters.

And lest you think I'm being smug & superior here, this isn't just hypothetical to me - I have computers that I don't run folding@home on.

I think community and user engagement is almost equally important (to me) than whether the folding is completed on time. Without users contributing to the science, the science won't progress.

Perhaps it would be interesting to see the breakdown of hardware that actually contributes currently to FAH? This would allow some sort of empirical basis for the arguments presented earlier (thanks to everyone for the enlightening debate).

But maybe we should be wary about getting too stuck into each other, if we're wanting to build an inclusive community (i.e. everyone), that is.

Ad in many cases what you propose here have been tried, found wanting, and F@h has moved on.

@Joe_H, I'm wondering if non-technical aspects have been thought about? As increased total performance could be done via both 1) increased code-efficiency and 2) increased user participation levels.

nchowlett wrote: ↑Tue Feb 28, 2023 11:56 pm I think community and user engagement is almost equally important (to me) than whether the folding is completed on time.

Perhaps the scientists who are running the project already know what's important to them. They're the ones doing the research, not you.

You are here as a volunteer to help them. They are not here to help you.

nchowlett wrote: ↑Wed Mar 01, 2023 12:09 am @Joe_H, I'm wondering if non-technical aspects have been thought about? As increased total performance could be done via both 1) increased code-efficiency and 2) increased user participation levels.

If the extra participants have the hardware needed to make a valuable contribution, them bring them on. But adding more participants just for the sake of making the user count go up isn't a worthwhile goal.

Perhaps the scientists who are running the project already know what's important to them.

Perhaps you should realise without the community they can't do squat.

If the extra participants have the hardware needed to make a valuable contribution, them bring them on

Agree. But what if the 'user experience' is the thing that's reducing the amount that make it through the door to contributing?