Techno Musings

Last week at SC09 IBM announced that a Bluegene/P system with 147,456 CPUs and 144 terabytes of main memory was able to simulate the cerebral cortex of a cat.  Personally I find this kind of work extremely exciting and really on the bleeding edge of computer science.  One of the examples of how the simulation was used comes from being able to view the waves of processing passing through the simulation to better understand the structure and behaviors of the feline brain.  I’ve taken an extract from the IBM press release about the overall SyNAPSE project sponsored by DARPA to better explain the context.

After the successful completion of Phase 0, IBM and its university partners were recently awarded $16.1M in additional funding from the Defense Advanced Research Projects Agency (DARPA) for Phase 1 of DARPA’s Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) initiative.  This phase of research will focus on the components, brain-like architecture and simulations to build a prototype chip. The long-term mission of IBM’s cognitive computing initiative is to discover and demonstrate the algorithms of the brain and deliver low-power, compact cognitive computers that approach mammalian-scale intelligence and use significantly less energy than today’s computing systems. The world-class team includes researchers from several of IBM’s worldwide research labs and scientists from Stanford University, University of Wisconsin-Madison, Cornell University, Columbia University Medical Center and University of California- Merced.

I cited this portion of the announcement letter because it points out a goal of SyNAPSE is to create a custom chip, basically an ASIC.  What this tells me is that on the bleeding edge of computer science to do things like simulate mammal’s brains, hybrid compute systems with special purpose ICs are required.  As I have talked about in the past with respect to Bluegene, it too uses both compute ASICs and communication ASICs as fundamental building blocks.  Note that there are numerous articles on the Internet which summarize the research and a fabulous blog on cognitive computing which I suggest that you check out.  My intention from here on out is not to rehash or summarize the work, that is already done, but instead to point out why you as an IT professional should care about this research.

Essentially there are two major reasons to care.  The first is a little altruistic because this kind of research done for DARPA is rather like the work within particle accelerators; namely, both are accelerating an increase in the general body of scientific knowledge.  In this case the acceleration is into the realm of learning and understanding how we and our cousin mammals think.  The potential applications of this kind of basic research are huge simply because the research is so fundamental. Things that are hinted as potential benefits are future systems that can recognize needles in the haystacks of information.  These are relatively simple things that an organic brain can do, but a computer has a hard time actually keeping up with.

Secondly, these super computing architectures are significant to computing platforms because they set the tone and direction of the technologies deployed in business 5-10 years in the future.  So basically our future super geek toys will be derived from the ideas, architectures, and artifacts that these advanced computational labs are dabbling in today.  Further, to achieve the goals which DARPA has set forth for SyNAPSE there have to be serious advances in dense green computing, storage, networking, etc.  After all if we can only create a super computer that thinks like a human if it is next to a nuclear power plant and larger or as large as the Superdome in New Orleans what is the point.

Personally, I continue to be happy to see the advance of hybrid computing systems and concepts.  In one of my more recent posts I explored a project from Microsoft R&D called Barrelfish that seeks to bring to the attention of the IT community the fact that there are many core platforms already in the wild and that we need to create OSes and concepts that can best make use of them.  Further Apple with Grand Central Dispatch also seeks to make it easier with their OS for programmers to best take advantage of many core systems.  Add in the Feline Computational simulation with Bluegene/P and I think we can predict that the future will be hybrid computing platforms for a variety of applications.  I can honestly say that I’m excited to see the next pointer in the future leading to the widespread deployment of these hybrid compute, storage, networking, etc. architectures.

UPDATE

There appears to be some controversy over the validity of the feline simulation.  Apparently a rival of the lead behind the cat simulation project has some doubts.