I visited the US from April 17 to 26 to attend the Program Committee meeting of the OOPSLA International Conference, to discuss with researchers in parallel C++ research groups in the universities listed above, and to see current nCUBE3 products (nCUBE Co.) and myrinet network products (Myricom Co.) I will talk about the universities and myrinet that I visited in this report.
The University of Indiana is located about 45 minutes south by car from the Indiana airport. With no direct flight from San Francisco, I went to Indianapolis by way of St. Louis. At the airport, I waited nearly 4 hours due to a delay in the departure. Although I knew I had to wait a bit longer for a connecting flight, I never expected that I had to wait such a long time.
In a rnorning rneeting at the University of Indiana, I discussed MPC++, currently being developed at Tsukuba Research Center. The afternoon was spent mainly hearing about MPC++ (aiming at a de fact ostandard) being developed by Professor Gannon's group. He said that HPC++ is now in the process of being standardized, mainly at the Universities of Indiana, Colorado, and Oregon, as well as at PGI, IBM and NASA Los Alamos.
I heard that standardization is also under way in Europe under the name of Europe C++. It seems to me that no one knows where the research and standardization of C++ parallel/distributed extension are heading. I left the University in the evening to fly from Indianapolis to Urbana-Champaign.
The following day (April 20), I presented a current status
of our research at the University of IIIinois, and had additional meetings
with Professors Chien, Agha Kale and Muroga. Fortunately for us, Professor
Chien is very interested in the application of MPC++ and RWC-1. He asked
me about MPC++ procedural matters so they can use it as soon a possible.
After these meetings, I left for San Francisco in the evening and checked
into a hotel in Palo Alto around 11:00 p.m. Before departing in the afternoon of April 22 (Tuesday), I
had a meeting with a person from Myricom Co. at a hotel near the Los Angeles
Airport. Myricom is a venture capital business founded by a CalTech graduate.
It is developing, manufacturing, and marketing network products called "myrinet."
More and more US universities and colleges are using myrinet to develop
workstation clusters for experiments with parallel systems. Myricom does not have promotional brochures, and projects
are introduced only via WWW. I heard that Myricom has no interest in doing
business with anyone who cannot access WWW. Asked if Myricom was interested
in making money, he replied that it was enough for their products to be
used in the academic world. He added on his leaving: "My yacht is moored
in the harbor over there." This convinced me of Myricom's business position.
We are planning to have an international workshop on an advanced
C++ parallel/distributed system sponsored by RWC in March, 1996. We asked
professors Gannon, Chien and Kesselman to be program committee members.