HPC Tech Trickle-Down

Print Friendly, PDF & Email

Tom Foremski of ZDNet posted an overview and evaluation of an article [really blog post] from Irving Wladawsky-Berger, former chief strategist at IBM.  For my next trick, I shall attempt to write an evaluation of Tom’s evaluation.  Given the recent buzz about the SC09 show floor, there have been a plethora of objective pieces posted as evaluations on the tech involved to get us to an exascale platform.  As the number of articles written increases exponentially, so do the variance of opinions on how we’re going to get there.

Another massive technology and architectural transition now looms for supercomputing and the IT industry in general.” [Irving Wladawsky-Berger]

That being said, Tom has taken somewhat of a different approach.  His article is really concerned with how the exascale technology will eventually trickle down into the realm of traditional IT.

If we can build exascale computing supercomputers, we will also be able to build smaller systems that use the same basic technologies and that will have many business applications, not just in cloud computing. [Tom Foremski]

After having recently attended the Hybrid Multicore Consortium meeting in San Francisco, I have somewhat of a new appreciation for what exascale might require in the way of technological leaps.  Unfortunately, many of these leaps in technology will require our fair industry to fundamentally change the way we currently do almost everything.  Core processing architectures, memory architectures, operating systems, programming models, algorithmic techniques and simple user interfaces are all on the chopping board.

If we can crack the formidable challenges of processors, software architectures, and how to develop applications designed for extreme parallel processing, we will be very close to cracking some very large computational problems that currently are out of reach, from “economics and medicine to business and government.” [Tom Foremski]

Enter fickle [yet cost conscious] Fortune 500 CTO, stage right.  I find it hard to believe that traditional IT departments are ready to upset their careful balance of stability, security, resilience and usability.  These fundamental ideals are the warm blanket that separates traditional IT from the bleeding edge psychosis that is supercomputing.  Agreed, IT will certainly see the eventual trickle down of the breakthroughs incurred as we seek out the next order in computational magnitude.  However, don’t hold your breath.

For those interested in traditional IT, I highly suggest reading Tom’s article.  It presents a compelling argument.  I also suggest reading Irving’s analysis.