I had a wonderful time in San Jose and got to see a lot of really amazing technology both in hardware in software. The high performance computer market is poised to genuinely explode in the not too distant future. In case you haven’t heard, that what Jen-Hsun Huang is actually brandishing in my previous post is a mock-up of a GT300 and not the real thing. It’s been something of a dust-up on the GPU sites. Personally, while I suppose I do feel a bit duped since I had spent considerable money to be sitting in that chair when I took the photograph, I really don’t care. The tech is real and is coming and will benefit the science. Besides, I’ve been to enough trade shows in support of my own company to know that sometimes things just aren’t quite ready in time for the show.
Perhaps the highlight of my trip was the early evening visit that I was able to pay to the Pande Labs. My friend John Van Arnam who is a post-doc student at Stanford was kind enough to show me around a little bit, introduce me to some of his friends and take me by the Labs. Afterwards we had sushi followed by gelati with his lovely and charming wife. What more can you ask for really?
Nitteo, a major folder and genuine pioneer of fold-farming on CPUs and GPUs for team OCN decided that he needed to take a break from folding and focus on his family and bussinesses in these troubled economic times. Of course what would happen to all of the equipment that made up his folding farm was a very big question.
I’m proud to announce that another folder (VanArnam of team Fold4Life) and I have shared the investment of purchasing the bulk of Nitteo’s folding farm: 31 9800GX2 dual-GPU graphics cards, 5 motherboards, CPUs and power supplies. This equipment approximately doubles the total folding power of Atlas Folder and will be integrated into the existing Atlas Folder folding farm May 7 – 10th.
I spent several hours last night unpackaging and inspecting the GPUs and made a short video to commemorate the event.
Today’s addition of a fourth full rack shelf with 4 GTX295 graphics cards installed has brought Atlas Folder to my initial build target!
19 GTX295s (38 GPUs)
1 GTX260
2 8800GTX
1 PS3
6 Quad Core SMP
34.219 single-precision TeraFLOPS
I am very pleased with the way that the project has gone overall, and if I had to do it all over again I can’t say that I would change very much. The machines have worked well, the various components that I’ve chosen have been pretty cost effective and stood up to the load, and most importantly the F@H Work Units just come flying out!
I greatly appreciate the support that I’ve received over the last few weeks from all of the readers of atlasfolding.com, but especially to my family without whom a project like this would be vastly more difficult.
Now I have a decision to make: whether to continue building Atlas Folder using GTX295 boards or whether to pause where I am and wait for nVidia’s 300 family… Opinions anyone?
In other news, I was recently accepted to the Folding@Home beta test team. Testing the new work units has caused a dip in my daily PPD, but I’m more interested in assisting the science than the pure points per day race. I’m very glad that my investment in Atlas Folder can be used to help hasten the delivery of new work units to the larger pool of contributors.
Today AtlasFolder will pass DiverGuy on my folding team HUNT-DIS to take the number one position.
While I’m proud of my contribution to Folding@Home I feel I have to take the time to thank DiverGuy and mwbrownibm on the HUNT-DIS team for their longstanding and important contributions to the cause. Both of these gentlemen and others on the HUNT-DIS team have been folding since before it was cool in order to reach their point totals and I want them to know that I sincerely appreciate all that they’ve done for the cause of Huntington’s Disease and Folding@Home in general.
Prof. Pande made a post in his blog a few days ago regarding the submission of a new paper on the Huntingtin protein entitled “The predicted structure of the headpiece of the Huntingtin protein and its implications for Huntington’s Disease” to the Journal of Molecular Biology by Nicholas W. Kelley, Xuhui Huang, Stephen Tam, Christoph Spiess, Judith Frydman and Vijay S. Pande. The abstract can be seen here. Prof. Pande has promised more information on the research in the future.
While the research itself is highly technical I do relish seeing feedback (especially relating to Huntington’s and CAG repeat disorders) from Folding@home contributions.
While browsing YouTube for Huntington’s and Folding@home related films I found this somewhat dry but nevertheless informative video of a protein folding. The film shows a folded protein molecule and then an unfolded string. It then proceeds to show the iterations that the protein molecule goes through searching for its final stable fold.
The video introduction says the folding occurs in 6 microseconds of time. 6 millionths of a second…
In other news, I was recently accepted to the Folding@Home beta test team. Testing the new work units has caused a dip in my daily PPD, but I’m more interested in assisting the science than the pure points per day race. I’m very glad that my investment in Atlas Folder can be used to help hasten the delivery of new work units to the larger pool of contributors.
