Futures In Biotech

[Jesse_V]

This may be something that only Dr. Pande can fix, I don't know:
Just now I went on folding.typepad.com and clicked on the link to watch the Futures in Biotech show. The link is http://twit.tv/fib85
It says the page does not exist. However, after some Googling I found this: http://www.twit.tv/show/futures-in-biotech/85 which looks like the same thing. I didn't watch it yet because the GPU client was making things super slow to respond on the page, but it should be the right show.

I suggest changing the link.
Thanks.

[7im]

Pinged PG.

[PantherX]

Weird, the link was working over the weekend so I got myself a copy.

[Jesse_V]

Yeah your right I noticed too that it was working just a bit earlier, but I didn't download the video or anything then. Rechecked it just now and its still not working.

Haven't watched the whole video yet, but from the beginning and some bits in the middle it was quite fascinating, very impressive, and well spoken. May I just say that in my opinion Dr. Pande has an excellent speaking style. Explaining something very complicated in simple language is a skill I think few have. Well done sir!

[7im]

Fixed first thing this morning.

[Jesse_V]

I see that! Perfect. Awesome.

[GreyWhiskers]

I watched the whole video last night. Really fascinating stuff. Promoting the upcoming v7 client looking for even easier use for new folders.

The insights into the molecular biology aspects of the project, and how they relate to the computations, was good to see.

A couple of interesting notes from Dr. Pande. He expressed how much work gets done on the GPUs, and envisioned that even 100K GPUs would be an enormous boost to the science being done. (the stats now are showing about 9700 active ATI GPUs and about 9,000 active Nvidia GPUs. There is a LOT of room to grow here).

One interesting question was asked about use of the CPUs in modern TVs for folding. That may be a new avenue to pursue.

The video runs about 1 hour 15 minutes, and is easily found by googling "futures in biotech".

[Jesse_V]

I watched the whole video last night. Really fascinating stuff. Promoting the upcoming v7 client looking for even easier use for new folders.

The insights into the molecular biology aspects of the project, and how they relate to the computations, was good to see.

A couple of interesting notes from Dr. Pande. He expressed how much work gets done on the GPUs, and envisioned that even 100K GPUs would be an enormous boost to the science being done. (the stats now are showing about 9700 active ATI GPUs and about 9,000 active Nvidia GPUs. There is a LOT of room to grow here).

One interesting question was asked about use of the CPUs in modern TVs for folding. That may be a new avenue to pursue.

The video runs about 1 hour 15 minutes, and is easily found by googling "futures in biotech".

On my machine the GPU less heat than the CPU, so I tend to run that more. Bear in mind that the summer heat may have a lot of influence on the number of folders. I remember back in December 2010 we were up at 10 x6 petaFLOPs overall, but I don't remember how many GPUs were participating at that time. Another thing to keep in mind is server or client issues. IMO, that probably isn't a big of a deal compared to the heat. Searching for a cure for Alzheimers and such is nice and all, but if things start overheating people will shut down. You are right though, we need to encourage others to fold however they can. The GPU is an important tool. I've read somewhere that every CPU/GPU/PS3 they gain increases their speed by a factor. So in other words if they have 100 GPUs, and gain 2 more, science moves along at 102 times as fast as a single GPU. They are that good at parallelizing the calculations!

As to the CPUs, I didn't watch that part yet, but I don't think they will go that route just yet. It may be similar to the reason that their not folding on the Xbox: not fast enough processor. Plus, its a TV, so its pretty commonly used. I don't know how TV CPUs manage things, but I would guess that it can only do one thing at a time. Meaning that F@h cannot be run in the background, and so would stop people's TV usage. Of course the PS3 can't running F@h in the background either, but gamers always have TV to fall back to. Those are just my thoughts about the whole thing.

You can find the video by Googling, but the link in folding.typepad.com works now, since they changed it to the link I suggested.

[Jonazz]

Very interesting indeed!

I was a bit surprised to hear how good he spoke about GPU's, and how important it is to get more GPU users. The GPU's might deliver more FLOPS than CPU's, but isn't getting more CPU's more important because GPU's can only handle small/easy proteins with few atoms?

In any case, it's great to hear how F@H has evolved over the years and that they are finally getting some results actually related to diseases. I can only dream about what wonderful things can be done in the future! Congratulations to all members of the Pande Group!

[bruce]

. . . because GPU's can only handle small/easy proteins with few atoms? . . .

I have not listened to "Futures..." yet so if he said something specific there, it will be more accurate than what I'm about to say... but I'll say it anyway. GPU technology is progressing very rapidly. The old limits of small/easy proteins with few atoms are rapidly giving way to bigger/harder proteins with many atoms. The early FahCore builds with limited versions of Gromacs are being updated to better/more powerful versions with more scientific options. This trend it likely to continue.

Older generation CPUs and older generation GPUs eventually get updated to newer generation hardware and in the process, the some of the older hardware needs to be deprecated (eventually). See the comments here about the deprecation of older ATI and older NV GPUs.

[Jesse_V]

Very interesting indeed!

I was a bit surprised to hear how good he spoke about GPU's, and how important it is to get more GPU users. The GPU's might deliver more FLOPS than CPU's, but isn't getting more CPU's more important because GPU's can only handle small/easy proteins with few atoms?

In any case, it's great to hear how F@H has evolved over the years and that they are finally getting some results actually related to diseases. I can only dream about what wonderful things can be done in the future! Congratulations to all members of the Pande Group!

It is my understanding that all clients are equally important and appreciated. The GPUs deliver more FLOPS because they're hardwired to quickly do calculations that would take a regular CPU much longer. If you read the PS3 FAQ on the F@h site, it mentions that GPUs do the fast calculations, but aren't very flexible in the types of work units it can do. The CPU is slower than the GPU, but is much more adaptable and can take just about anything. I think the PS3 is more flexible than the GPU, but isn't as fast for certain calculations. So each has its pros and cons I guess. I see no reason why GPUs can't take on large proteins, but I guess it depends on how much memory your GPU has. One of the GPU WUs I just finished (6601 (Run 9, Clone 104, Gen 163)) has 1047 atoms.

I am very impressed as well. Some of the recent papers are just incredible. In my opinion, if there's any group who's going to cure these disease utilizing simulations, it will undoubtedly be Folding@home. And we're all a part! I feel so skilled or something...

[GreyWhiskers]

I've had a series of these WUs assigned to my uniprocessor. I haven't seen any scientific results yet, but Dr. Izaguirre's info from earlier in the summer point to exciting possibilities for GPU folding.

i found this post by Dr. Izaguirre quite promising - especially that the method for potentially accelerating the MD computations by 100 fold being applied to the OpenMM GPU Core - with the GFlops capability of relatively cheap Nvidia GPUs, like the 560Ti, being north of 1 TFlops! As Dr Izaguirre says things are looking good! Stay tuned for an exciting summer (in northern hemisphere).

Re: Projects 10012-10085 -> accelerate F@H 100 times?
by izaguirr » Sat Jul 02, 2011 12:27 pm
...
For those of you who are curious, by the end of the summer 2011 we hope to have characterized the performance / accuracy tradeoff of this methodology. Things are looking good. Then we hope to deploy it for select projects on an OpenMM GPU core by the end of 2011. That means that the methodology will be at first restricted to the types of simulations that can use GPUs.

A second stage would include extending the methodology to work with general simulations that can be then incorporated into core A4 (or its successor). We plan on doing this in 2012.

Even at its most successful though, different scientific projects require different levels of modeling fidelity/accuracy. Thus, while we hope that this methodology can enable projects spanning much longer scientific timescales and eventually much larger systems, don't expect the standard CPU and GPU cores to go away ever.
Prof. Jesus A. Izaguirre, Ph.D.
Department of Computer Science and Engineering
Interdisciplinary Center for Network Science and Applications
University of Notre Dame, IN, USA
Member, Folding@Home Consortium, Stanford University, CA, USA

Folding@home project descriptions

Project 10012

This project is a test of a novel means to significantly accelearate Folding@home Molecular Dynamics calculations by 100x without any additional increase in hardware. Thus, this method has great promise to push Folding@home way beyond what it can do now, although there is much work to do to test it to make sure that the results are scientifically valid and useful. You can find more information about the method by reading one of our recent papers.

MULTISCALE DYNAMICS OF MACROMOLECULES USING NORMAL MODE LANGEVIN