FS3 Meeting: 18 Mar 2007, Milwaukee

On 18 Mar 2007, ten people gathered at Bucketworks in Milwaukee to discuss the FireSeed Streaming Supercomputer (FS3) project.

The meeting started with introductions so everyone at the meeting would know a little about each other. Following intros, the project background and work to-date on the project were explained. This was followed by discussion of various technical and organizational aspects of the project.

New things that came out of the meeting include:
  1. Interest in the project from two new people, John G and Lars O. This interest gives the project some good ties to Milwaukee universities and more advisors for funding sources relevant to FS3.
  2. Knowledge about and potential contacts in Milwaukee-area CPU-cluster supercomputers.
  3. Suggestions to look at verynear-future work on learning CUDA on one machine and to consider whether two box/four GPU systems are powerful enough for some HPC needs (high performance computing).
A slightly disappointing aspect of the meeting was the low turnout, especially of people new to the project. There are no doubt 20 to 200 people in Milwaukee who would have a high interest in the FS3 project. The challenge facing the FS3 project is to connect with those people and let them know about the project. I've got several ideas along those lines and will develop and prioritize them in terms of effectiveness. Lots of ideas and not enough time!

A very encouraging aspect of the meeting was the high level of interest on the part of the people at the meeting who are new to the project. Both John and Lars voiced strong interest in participating in the project and identified FS3 areas of personal passion. In another example of the six degrees principle, when John started talking about potential users and applications, he mentioned Backlot Imaging because he knows some of the people who work there. Backlot Imaging happens to be listed as one of the FS3 project supporters on the grant questionnaire which is being used to pursue project funding. It is, indeed, a small world -- and this project should help to make visible more of the connections in the Milwaukee, Wisconsin and Midwest tech community.

Next steps for the FS3 project include scheduling an FS3 project intro meeting in Madison. Once the meeting date is set, it will be announced on both this blog and the project Google Group.

(Sparked by the meeting comments about getting more people familiar with CUDA, during the trip back to Appleton after the meeting a decision was made to pull together costs for two options to have online access to CUDA machines. Option one will be for a single machine with one NVIDIA 8800 card and option two will be for two machines which each have two NVIDIA 8800s. Once the costs and technical details for these two options are defined, the options will be presented to the FS3 project team to determine if funding should be sought for one or both of those options. Preliminary costs for option one appear to be in the range of $2000 - $2500.)