21st Century postmortem

About a week has passed, and a friend said something quite interesting about this. Michigan has decided not to invest in Michigan’s future. Amusing.

The commercialization areas that we submitted for were to be scored on Scientific and Technical basis, Personnel, Commercialization Merit, an fund leverage. The first sentance of the review sets the tone. “The technical merit of this proposal could be phenomenal, but there is a lack of explanation of the important underlying science.” Interesting. This is a commercialization, of a technology. The underlying science behind the technology is already very well understood. The application of the technology to the scientific field we used as our model is also extraordinarily well understood. From there “It appears that Dr. Landman seeks to combine all of the tools of the current state of the art related to ???floating point processing systems??? to build a system that is faster and more effective that current computational systems. Dr. Landman speaks of developing systems that are orders of magnitude faster and significantly more cost effective than current computational systems." Yes. Exactly. Precisely. On the button. And as we indicated, throughout this entire proposal, this applies to far more than just chemistry, building a faster/better molecular dynamics engine, but to engineering computing, and … Well, you get the idea. We pointed out in the proposal that we planned to expand this same system into engineering computing. Into other areas that would be well served by extremely high performance systems at reasonable prices. The subsequent two lines show two fundamental misunderstandings: 1 of a market, 1 of the application base. “The only company that is managing to keep going in this area is Peta Computing, makers of the MD-GRAPE system, and that is a very different and highly specialized type of accelerator system specialized for one particular type of problem (molecular dynamics). We just don’t think there is any evidence that a commercial entity can survive marketing special accelerated biocomputing hardware." We were talking about a more general system than Grape. Specifically we pointed out Grape was a highly specialized bit of hardware good for precisely one thing, and that one thing did not make it well suited to do the sorts of things we had in mind. We pointed out that we planned to work on porting LS-Dyna, and other non-bio/non-chemistry codes to our systems. This would be effectively impossible for Grape. Moreover, they say that they do not believe that there is any evidence for success in marketing accelerated biocomputing hardware. I have pointed out many times that clusters are a form of accelerated hardware. Many more cycles, and they are wildly successful in the bio space, and actually in all spaces. In a bigger sense, accelerated computing is a massive market today. Almost everyone owns a supercomputer. In their GPU. Again, highly specialized ASIC silicon, but it computes graphics calculations so quickly that some people are willing to rewrite code to run in its highly restricted execution model to get better performance. The market is there. It has been waiting for a product at a reasonable price point. No one in their right mind is going to pay 100k$ for an Orion 96 CPU system (and we made that point to the VCs a few years ago when they were considering funding them). Offer a system on a board that can get you 10-100x the performance on a small range of critical apps, and provide an API, and you will find a much larger market. That was our point. Molecular dynamics was the low hanging open source fruit we would work with. Another curious aspect of this was our match. I won’t talk about who/what, only how much. We had in excess of $6M of matched resources. We were told this was weak. Uh….er …. ummmm….. There are many colorful expressions I can use to express my bewilderment. I found one on wikipedia, but I am not going to link to it here. Michigan just missed an opportunity to invest in Michigan’s future. My friend is right. At the MGCS, the chief economist of Comerica bank painted what I would call a doom-and-gloom forecast for the local area. The take away message was, unless you start working on a change from manufacturing to a knowledge based local economy, you are going to get left behind. The 21st century fund SEIC board just voted in favor of the manufacturing economy. Update: Note added in proof. SGI, Cray, Sun, etc are falling all over themselves to get FPGA based accelerators into the market. So are a few others which we cannot mention. NEC just announced a product introduction in the US in a few months that is very similar to what we had proposed in our application. Imagine that. No market. Thats why everyone is chasing this null market. Update 2: some weeks later Into this null market, more and more companies rush. Read announcements of plans and products for the past several weeks. Words fail me.