PS4 / XB1?

[bruce]

The psychology of folding on game consoles is unlikely to have changed. For reasons I've already explained, in the long term, Stanford's portion of the investment in the PS3 client needed to produce more results. People would fold a few weeks and then either abandon it or move to a PC. Also, Moore's Law (which works for PCs) doesn't work for game consoles (upgrading to a bigger disk doesn't make it run faster). The grand total of game console folding hours per week didn't grow as fast as was expected. An outstanding viewer doesn't keep people interested.

[Jonazz]

Sony only put resources in developing the PS3 client to gain attention and sell more consoles. Appearantly, the benefits weren't large enough, otherwise they would surely be helping on a PS4 client.

But do remember one thing. The PS3 client was revolutionary at its launch. The cell had immense processing power. But a few years later the PS3's processing power was already outdated. Is it worth it (for Stanford) to spend so much time andm oney in a client that will be outdated after 2-3 years?

[NookieBandit]

Some fairly recent news on the PS4 looks like it has achieved 10 million units sold (http://www.extremetech.com/gaming/18793 ... -5-million). Not too shabby.

I have to believe that if a PS4 client were made available, Stanford would see it as a valuable contribution, if any comparison can be made to the value contributed by the PS3: The Wikipedia article on the PS3's contribution to FAH says, "Over its lifetime of five years and 7 months, more than 15 million users contributed over 100 million hours of computation to Folding@home, greatly assisting the project with disease research". I think a reasonable argument can be made for valuing the contributions in the first few years of the PS3's existence, given as Bruce points out, the seemingly unstoppable progression of Moore's Law. Perhaps the same argument could be made for a PS4 client.

I got my start folding on a PS3 in 2009. As I researched it further, I became a certified folding junkie and am now addicted to racks and stacks and slots and watts as my drug of choice. Folding on a console was a very nice way to introduce the concept of protein folding to the general public, and as Bruce says, some move on to using PCs (as I did) for folding in a much more serious way, and in my case went from two PS3's back then to 10 CPUs hosting 26 GPUs in use today folding 24 x 7. Basically, consoles are the the gateway drug; I have to believe they can be credited for boosting the ranks of dedicated folders, but I do recognize the limitations of one-person market research.

By the end of 2013, there were about 80 million PS3's sold, so the penetration of FAH on the PS3 installed base was very impressive at almost 19%. Assuming just half that penetration on 10 million PS4's sold to-date, each with about 1.8 TFLOPs of performance, that would add a total of a potential 1,710 petaFLOPS to the FAH processing network. That's pretty darn impressive.

I can see two ways of getting access to the installed base: Hiring a development team and funding a basic FAH client with a fancy viewer through private donations using a crowd-funding platform (http://www.crowdfunding.com/). The development team would have to sign all the obligatory NDA forms Stanford requires and comply with any other trade-secret concerns. Importantly, there is no need to have Sony directly in the middle of the development process, since the FAH client would be written using tools provided by Sony's Independent Developer Program, which it strongly supports (virtually free development kits).

The second would be more involved, but may also be more interesting. The funding concept is similar to above, but the end-product would be more in keeping with Sony's strategy of developing games that help sell more consoles. Watching my kids play first-person shooter and strategy games, all the rules and elements are present in protein folding to make it into a pretty compelling game. Ideally, the FAH Game could work mostly like a strategy game. The typical "world" in a strategy game would be in the FAH Game the "Energy Landscape", gamers head down different trajectories where the MSM begins to form structures, or limitations, around the movements a gamer can take or make. Events lead to pathways and vice versa, changing dynamically as the gamer makes different choices. Points accrue as the gamer is able to fold a given protein in the most efficient way (fewest moves) the fastest. Adaptive sampling would give every gamer a new beginning state to work from, which would open up the number of possible end-state folded proteins. The idea is to use some form of human intelligence in the protein folding process, but mask it in the form of a strategy game, and winning by finding the most energetically favorable conformation of the protein. Points get taken away when a protein misfolds, and are earned back when a gamer uses embedded cellular rules to try to refold it correctly. The gamer loses (and FAH wins) when a protein gets misfolded and can't be refolded, aggregates and has to be destroyed. Substituting characters for each functional element of the game, creating a cool 3d Energy Landscape, making the challenges increasingly difficult over time and letting kids with gaming skills try to "win" by using rules that follow basic biologic processes would be awfully interesting interaction to watch and play. It might even spawn a computational biologist career or two, or a thousand. The FAH Game would be funded the same way as the first, but would need a more comprehensive story-board laying-out of the game with a higher funding goal. Of course, all that is dependent on whether or not human intelligence could play a role in advancing the speed and efficiency with which protein folding can be accomplished, so if I'm off-base by orders of magnitude, then my apologies for the long post.

At minimum, how does one find out if Stanford would be supportive of the first alternative (a simple FAH client)? And by "supportive", I don't mean monetarily, but just facilitate the process of ensuring a development team had what they needed (and what Stanford needs) to get the job done?

[ChristianVirtual]

I did not dig deeper but the descriptions on the front page to register for PS4 SDK looks rather liberal. The show stopper might come some level down the registration process (like NDA or overall capabilities).

http://us.playstation.com/develop/

https://www.companyregistration.playsta ... egion=scea
There are even flags for "university and academia"

And ocores might be the ideal methodology to deal with the potentially short power cycle times of consoles.

[ChristianVirtual]

Found this one: but it's not official, so just for your entertainment.

http://www.extremetech.com/extreme/1713 ... -gpu-specs

Since it's highly customized AMD-based chip a port might be not that easy and also not sure if the SDK would give access to the GPU in the same way as we have on current used hardware and software.


The concept from NB to design a game around folding and include the human brain is a creative one. But I could not fold 24/7 anymore

[bruce]

If you're interested in a video game rather than in scientific research, consider "Foldit" at http://fold.it/

As far as I know, it runs on Windows/Linux/MacOS and probably not on the game consoles. It can be very entertaining.

It was developed by the folks at UW and and it does teach you some of the complexities of the folding process. The objective is to find the shape of native state of an unknown protein. [FAHs objective is less aimed at discovering the native state and more focused on discovering and understanding pathways to mis-folded proteins.]

[NookieBandit]

Thanks Bruce, I didn't know about Foldit, but a quick overview looks like it does have some fun and interesting characteristics of combining human intelligence with folding. I'll have to look more closely at it over the weekend. Wifi on a plane makes app downloads problematic.

Do you have any suggestions on who I can contact at PG/Stanford if I wanted to pursue the idea of a PS4 FAH client? I could line up a few friends/fellow angel investors and get rolling on a product design spec if I can get some answers to the obvious questions of intellectual property licensing/protection (both ways), NDA requirements, non-competes, etc. After that, and with a relatively detailed spec, I'd be happy to manage a crowd-funding campaign. I'm an SEC accredited investor and managing director of a private equity investment firm. I have in-house legal counsel I could tap on to ensure everything is run above-board and transparent. Game development isn't an area of investment for my firm, so there would be no conflicts of interest in my role as a professional investor and my outside interest in FAH.

ChristianVirtual, following your posts, you clearly have a very good technical understanding and look at problem resolution very pragmatically. I think you could add significantly to scoping the challenges of a project like this and helping us understand the level of effort needed going forward. Any interest in participating, assuming the ground work with Stanford/PG can be accomplished successfully?

[7im]

Snip.

Do you have any suggestions on who I can contact at PG/Stanford if I wanted to pursue the idea of a PS4 FAH client?

There is only one. pande at Stanford dot edu

[AlexF@H]

I can tell you 15,000,000 + 2 reasons to make a PS4 and an XB1 F@H client:

1 - 10,000,000) PS4!
10,000,001 - 15,000,000) XB1! again...
15,000,001) OK, enough with the funnies! Let's get serious. The reason why computers were made is because they wanted to solve mathematical problems and not to entertain people. By this I mean that, since they have found a way to fold proteins and do stuff the program does in order to find cures about deadly diseases, the best thing you can do is to sacrifise some of your time to write the F@H program for some of the most famous and most productive computer systems (XB1 & PS4) and encourage people to do this like I am doing now with my words.
15,000,002) Also, these machines are very powerful like I and some others said before me. The servers must give them the most demanding, the most gigantic and with the biggest priority projects!

[3D_Now!!]

NookieBandit, you seem very keen to get a client rolling for these consoles, and you seem like the person who has the means to get it going. When the PS3 client was released, it was the largest contributor by FLOPS for a very long time, simply due to the power/machine and the large install base. There is no reason to believe that the PS4/Xbox One would be any different. I think predicting 1.7 exaflops is probably a bit excessive ( ) but it will no doubt be a massive increase in output which will accelerate the project forward. Which is what we all want, or we wouldn't contribute. Therefore, I urge you not to give up on this until all avenues have been exhausted, as you have the resources at hand and the enthusiasm to make this work. I would support a kick starter campaign for it, and I would also get all my friends to sign a petition which could be put forward to Sony/Microsoft.

I think the main way to sell this to those two companies is to go down the altruistic/PR route. Lots of companies do things with no direct financial motive (internal CSR which customers never find out about) so I don't think there is any reason to believe Sony or Microsoft wouldn't support this because it won't be profitable. Especially if the potential benefits to f@h are made clear (it would certainly get the geeks in the development office excited, there may even be some f@h contributors there already!).

If you can produce a credible case for them I think this can become a reality!

[VijayPande]

We are actively pursuing consoles, as we did in the previous generation. I can't say more due to NDA's -- these companies get very upset about this. They tell me that they get tons of calls, etc and it drives their support people nuts to have to deal with leaks and rumors.

I wish we could talk about everything going on. If you guys can get these companies to change their policies, then of course I'd be happy to talk about it. But don't hold your breath there.

[Brazos]

I have feeling this is at the bottom of their (Sony, MS) to do list. Just my hunch, nothing else.

[MrGoodBytes]

There was an aspect of this conversation I was really expecting to see discussed here that appears absent.... Sony has an agenda. They obviously spent a lot of money in development of the Cell processor for the PS3 and the goal was definitely to find more applications for the cpu than just video game systems. There is definitely a powerful statement achieved by the overall numbers they achieved to the research community. (and they had some similar foray attempts with the PS2, I believe they had a linked cluster at one time that was on the top500 list a very long time ago). Their current design direction is nothing like the past. Back to their agenda, I see a lot of bean counters pulling their hair out on warranting systems where there are people pushing their hardware full tilt 24/7 rather than a few hours a week. I cant see them wanting to incur that financial risk, legal risk (it would only take a few home fires), or PR backlash for increased system failure rates. Just my 2 cents.

[bruce]

While you're probably correct about their agenda, engineers design computers and (supposedly) make informed decisions about design alternatives. Part of the design process includes designing for worst-case scenarios, and if the system is unable to dissipate the heat generated by (for example) a CPU load of 100% in an extremely hot room after a certain amount of dust accumulates in the heat-sink, the system needs a better thermal design.

If an atomic power plant has a meltdown because the emergency generators needed to power the cooling system are in a basement that can be flooded, it's a defective design. Of course the engineer would need to have considered the concurrent possibility of both an earthquake and a Tsunami.

Designers are human, but the probability of CPUs at 100% 24/7 is much more likely and the lawyers will have to sort out if it should be the responsibility of of the owner of the CPU to manage the heat or if the design was inadequate.

[MrGoodBytes]

well let me say Sony's past initiatives in alternate "uses" was very influential on me, I took a bunch of MIPS assembler courses in college and having the PS2 linux kit at the time, it was a really big revelation that I could run my assemblies natively rather than simulated and it caused me to dive into that kinda work whole-heartedly.... (I also own a PS1 net-yaroze).

but about risks, yes you assume there are some critical temp shut down functionality and such but for the general population, for people that have systems laying on shag rugs or crammed/stacked in entertainment-systems/bookcases with all vents blocked or the dorm mate who is building a laundry pile on top. they can't control this and it sucks. History of the last few generations on both sides have certainly been in the media for failure rates. What is the percentage of the population that knows the term: Red Ring of Death? And as a result both companies have seemed to be rather flexible on warranty lengths (I've personally had a system replaced more than 3 years after purchase without hassle). There are definitely bean counters at work on that decision. I can fully see their side of this that offering this kind of functionality has zero impact on actual sales or market share and negligible impact to PR

Sony is not in the same stable financial boat that they were in 10 years ago. The prior two generations, you could argue that Sony had an open market potential of adapting systems for research computational systems due to their unique nature of their hardware design. Those can be some very lucrative contracts if you can garner the interest. I don't see any of that in their current specs that could justify the decision for exploring that vs off-the-shelf parts. Straight up, Sony is in survival mode right now and they are operating lean and history did not pan out on paid research cluster front. Sorry I think it's just the hard truth.

Sorry I'm pretty sure I've pushed a few people buttons saying this, but it is the devils advocate point of view and I don't think the conversation is complete without it. Pick your battles on how you want to approach it. It sounds like there has at least been talks and I don't hear a hard no. I have always been on the Sony end of the battle but a big proponent of that is because the previous three playstations have all offered home-dev kits or ways of running my own code in some manner. Yes, I am a tiny sliver of the demographic but for me that's huge.... and I am a microsoft developer. I have NOT bought a next gen system yet, I haven't made up my mind and per history my decision rests on value added for an investment from someone who is a tech enthusiast but not really a gamer. [that's my loaded statement that I am a big proponent on seeing this being available]