ARM coming to our home computers, F@H possible? Also, PS4.

[Decembermouse]

http://www.tomshardware.com/news/Tudor- ... 12889.html

And many other sources confirm that Windows 8 will run on ARM. I wouldn't be caught dead with ARM on my main computer, unless it's cross-compatible with x86 and I just don't know how that's going to go. Different architecture and all. Weaker per MHz, compared to x86 CPU's, is what they always say. Hence why ARM goes into phones and tablets, because they are less capable processors, so they don't eat up as much power or produce as much heat, so they're realistic in tiny or ultramobile devices.

But with ARM looking to take over a significant portion of the home computing world, I find myself wondering if Folding@Home will be left behind, or if Pande Group will get on that train and make use of Folders who get an ARM rig. I realize cores and code, etc., would have to be largely rewritten and it would be a huge hassle. Whereas with GPU's and PS3, they took the plunge, that was understandable- as these platforms provide huge amounts of computing power.

But would ARM be worth it for the Folding software engineers to port it over? Or will the hardware be just too incompatible, or too weak, to make it worthwhile?


Also, I saw that the next-gen Wii (Wii U iirc) and the PS4 are taped out, and am excited to see if the PS4 also will run Folding@Home. The Wii U will have a tri-core IBM PowerPC processor, but I haven't seen any specs on it yet. Perhaps it could run "-smp 3" but then again the Wii culture isn't exactly known for being tech-savvy. I'd hazard a guess that a greater fraction of PS3 owners are Folders than 360, and that 360 also > Wii. But the PS4 should be technically capable, so I hope Sony works with Stanford again on this!

[7im]

No other CPUs will work for FAH without a powerful FPU. ARM chips run apps well, but they don't fold well. Mostly the same reason people aren't folding on their cell phones.

This has been well discussed already.

Fah on mobile devices (ARM)? [No]
Fah on an iPhone? [No]
Fah on an iPad? [No]

There are more, Fah on XBox, Fah on Wii, Fah on Atari 2600. Forum search is your friend.

[MtM]

http://fahwiki.net/index.php/Alternative_FAH_Platforms

Btw afaik those three cores are all smt capable so six threads.

Also, Pande Group uses mostly code from other initiatives ( gromacs/openMM/openCL ), if you could get gromacs to make a core work on the wi u I think f@h wouldn't be far behind, but I also don't think you're expecting them to write their own code specifically for the new Nintendo nor ARM ( see above link )?

Also, no technical details are available for the ps4 afaik so I'm not sure how you can be so sure that the ps4 will be technically capable? Who knows they might use another cpu architecture and a totally new client would be needed to run on it, even wurse while it might be capable, it could also be not capable though I expect Sony's pr department being savvy enough to see this connection has only given them positive publicity.

[bruce]

The goal of ARM (and the Atom and "Desna" Fusion SoC) is to maximize battery life by removing any functionality that's not needed in the netbook/iphone market. That's a diametrically opposite goal of FAH, which is to use as much of the rarely-used functionality of the PC/GPU as possible.

Folding on battery power is essentially impossible and finding unused TFLOPs on an ARM device is essentially impossible, too.

When plugged into the wall, my Atom-based netbook earns about 100 PPD. When running on batteries, they go dead in about 45 minutes.

I have no doubt that there will be a small overlap, such as the successor to my netbook but if anything, they're likely to move even further apart rather than closer together because of opposing design goals.

[MtM]

If ARM is moving to servers and they are, sure servers are more even then desktop very focused on power/performance but this still indicates that ARM in X time will be able to offer TFLOP performance to keep the nomenclature the same ( but not used as logical measurement, just objective ).

I think in 10 years ARM will be more prevalent then now, much more. Windows 8 native ARM support will help for sure. And not all these devices will be portable, I do see ARM coming to the desktop as well. I think ARM will undergo much more segmentation as we will have the same segmentation we now have with x86, low power netbook to high power quad socket server.

But, that aside, PG isn't going to provide a core for it, though maybe GROMACS will if/when ARM shows it has the performance needed to be a viable platform.

[Decembermouse]

This has been well discussed already.

Fah on mobile devices (ARM)? [No]
Fah on an iPhone? [No]
Fah on an iPad? [No]

There are more, Fah on XBox, Fah on Wii, Fah on Atari 2600. Forum search is your friend.

I know, I know... I've seen 'em all. I'm aware of these threads and understand why each of those things (except Atari 2600, I'm surprised anyone suggested that ) are just unrealistic. I already know that Folding on phones is useless to even speculate about. I was trying to bring up ARM in a different light with this thread, since we will soon be seeing much more powerful variants of the technology than you carry around in your pocket- desktop computers running with ARM chips. I thought the thread heading would have conveyed that I wasn't talking about iPad, phones, etc. In the same way an x86 CPU low-power enough to run a phone for 48 hours wouldn't be powerful enough to get anything meaningful done, and in fact those don't really exist anyway, so it goes for ARM chips in phones. This is old news to me...

Again, the difference here, and what I was trying to discuss, is ARM in the PC environment- multiple cores, probably quad-core or greater, higher clocks and larger cache, and an OS like Windows 8 that can run on said chips... THIS environment seems like the right stage for Folding... unlike in a phone (which has been talked to death and I really didn't want to bring that up here) ARM in this incarnation actually sounds like it might be realistic for running Folding. that's why I'm bringing the whole "ARM Folding" thing again, but with the "home computers" qualifier.

But if the FPU really will be that weak even on desktop iterations of the ARM architecture, then maybe it's not worth translating the Folding software over to be able to run on ARM. That is what I was asking.

http://fahwiki.net/index.php/Alternative_FAH_Platforms
Also, no technical details are available for the ps4 afaik so I'm not sure how you can be so sure that the ps4 will be technically capable? Who knows they might use another cpu architecture and a totally new client would be needed to run on it, even wurse while it might be capable, it could also be not capable though I expect Sony's pr department being savvy enough to see this connection has only given them positive publicity.

Yeah, I hope to see some more details on the hardware they've picked for the console, but nothing yet. I'm open to speculation such as yours though

[bruce]

In the smart phone field, ARM may be superior to what they're running now, and Win8 may be superior to Android or whatever else they're running now, but if we exclude that marketplace from the discussion, what is left?

Maybe your question, then, is will there be an X86+ARM (or X64+ARM) machine, and if so, how will it compare to an X86-only machine -- and will FAH even care.

I didn't mention the PPC, as seen on the older Mac platforms or the PS3. If ARM brings a plug-in-the-wall replacement machine like the PS4 which can improve upon the performance of the PS3 client then maybe. In fact, from the perspective of the science that gets completed, the current PS3 client is pretty much equivalent to the GPUs that were around when it was new. GPUs have progressed rather rapidly since then and it's not clear to me that a PS4 will be as big a step forward as the GPUs have undergone during the same time period, no matter
what technology is in the PS4.

[Decembermouse]

bruce, that's a good way to put it in terms of the PS4. The PS3 isn't any longer something I view as a good investment as far as science that gets done per Watt, and the PS4 may well end up in the same boat shortly after it's released.

With ARM, yes I think the ARM chips that will go in PC's will be superior to what they have now, but I'm curious whether they may end up with a powerful enough FPU, that without the help of an x86 chip next to it, to get work done. Right now, there is no market for ARM personal computers, but I find myself wondering: once they start popping up, and turn out to be surprisingly capable of running our day-to-day software, will they turn some heads around here, and prove themselves powerful enough, or a permanent enough component of the PC industry, to invest the time necessary to rewrite the Folding software for?

Maybe FAH won't care, that's occurred to me too... I wouldn't be surprised I guess, as it'd be more work than I can likely fathom. Best-case scenario for ARM, it would probably still take a long time for the code to get written, and first of course the ARM computers would have to prove they're powerful enough, and not just a fad that'll die out in a couple years.