|
|
|
|
|
|
|
|
|
|
|
|
Today, the dominant trend in high-speed computing architecture is to cluster several shared memory systems (nodes of the cluster) -- both vector and microprocessor. Well-established multitasking techniques are used for programming within a node, and message passing is used for communication among nodes. Clusters make it theoretically possible to apply thousands processors in parallel to a single application. Such systems have the potential to sustain trillions of arithmetic operations per second (TeraFLOPS) and are likely to play a major role in atmospheric modeling. SCD is well positioned to exploit clusters of vector systems and/or microprocessor systems.
Another significant development in FY1998 was a joint effort led by the NSF and the DOE to establish a Strategic Simulation Initiative (SSI) that could lead to placement of TeraFLOPS-capable systems in a few U.S. computing centers. SCD drafted a proposal for an NSF Earth Systems Collaboratory that could be a part of the SSI.
| Finally, after eleven years as Director of SCD, I elected to retire and go into consulting. An international search was conducted for my successor, and Al Kellie will become the SCD Director in early FY1999. I extend my best wishes to Al and to SCD. I’m sure that the next ten years in high performance computing will be at least as exciting as the last ten! | 
|
|
|
|
|
|
|