Cluster Computing

Once upon a time, I developed the StarT-X and StarT-Jr parallel systems. In 1998, a Hyades StarT-X Cluster with 32 400MHz Pentium-II Xeon processors was delivered to MIT's Earth Atmosphere and Planetary Science Department. For several years, the cluster was their primary facility for running MITgcmUV (a general circulation model for climate simulation) in research and courses.

  • A Personal Supercomputer for Climate Research. James C. Hoe, Chris Hill and Alistair Adcroft. Supercomputing Conference (SC), November 1999. (pdf)
  • MPI-StarT: Delivering Network Performance to Numerical Applications. Parry Husbands and James C. Hoe. Supercomputing Conference (SC), November 1998. SC'98 Best Architecture Paper. (html)
  • StarT-X: A One-Man-Year Exercise in Network Interface Engineering. James C. Hoe. Hot Interconnects VI, August 1998. (pdf)
  • StarT-JR: A Parallel System from Commodity Technology. James C. Hoe and Mike Ehrlich. Transputer/Occam International Conference, November 1996. (pdf)
  • StarT-NG: Delivering Seamless Parallel Computing. Derek Chiou, Boon S. Ang, Arvind, Michael J. Beckerle, Andy Boughton, Robert Greiner, James E. Hicks and James C. Hoe. Euro-Par, August 1995. (pdf)
  • Network Interface for Message Passing Parallel Computation on a Workstation Cluster. James C. Hoe. Hot Interconnects II, August 1994. (pdf)


A StarT-X PCI NIU Card


MIT EAPS Hyades Cluster of 16 2-way x86 nodes


MIT LCS Xolas Cluster of 9 SUN E5000 SMP's with 8 processors each


Two StarT-X Cards in a SUN Enterprise PCI I/O Board


Arctic cables (gray ribbons about $100 each) exiting the back panel of a SUN E5000 with 4 StarT-X cards


Four Arctic Test Boards (a 4-way switch)


Production Arctic Switch (populated for 32 endpoints)