Has Everyone Left Yet?

The last two instances of the HPCC Newport conference fostered the dialogue of an HPC reformation.  We started out looking at the level of HPC innovation and discussed various barriers to it.  We discussed how risk-laden a high-end system design win could be in these times.  And we looked at driving an agenda for reformation, ending in a call for participation in the reformation.

During this past year, the continued assessment of the HPC environment has brought into focus that much of what we’ve been looking at is symptomatic of a larger challenge lurking behind the scenes. It has become apparent that the HPC market segment has been characterized as prematurely aged. As such, the wherewithal to encourage innovation for the segment’s ongoing performance requirements is subjected to a narrow tunnel vision. The questions that have been raised are of the sort: who are we all comprised of? Who is being left behind? What are the implications? The recent dialogue has identified the existence of a relatively large un-served element within the possible HPC community, that which has been described as the “missing middle”.  These are users at the mid range of scientific computing for whom we must lower the bar for entry into modestly parallel computing. The “missing middle” concept is not new; it has been previously discussed in no less than the context of the Blue Collar Computing Initiative, created by Dr. Stan Ahalt, now at RENCI, and represents a large possible addition to our community.  Further, it has had an initial characterization performed by IDC as commissioned by DARPA, USC, ISI, and the Council on Competitiveness.

In this talk, I will review the progress of the envisioned reformation.  I will more fully define and describe the missing middle and what it might be like if the missing middle weren’t “missing”.  The motivation herein is to inspire further thinking and participation in the HPC reformation with an industry valuation of the role of facilitating the missing middle will have on what we have historically viewed as the supercomputer segment.  By enabling the missing middle to leave the starting blocks, there will be more energy for our journey.  We won’t ever get “there”, but we’ll certainly go faster and accomplish greater feats by having accomplished no one having been left behind.  Updates on Intel HPC technologies and activities will be reviewed throughout the talk, at the pertinent times.

We are at an inflection point; come and be part of history in the making.