The SLALOM benchmark has been described by Dr. Gustafson as follows
"Measuring performance over a fixed amount of time, instead of for a fixed-size task, had proved to be a good idea. But all the technical computing benchmarks did things the old way... testing time reduction for a given problem. So with the help of a newly-minted ace post-doc, Diane Rover, we devised the first scientific benchmark to ask how much work you can get done in a given amount of time... in this case, one minute. The parallel processor vendors were hungry for something like this, so it wasn't long before we had over a hundred widely-varied computer architectures on our list. (In contrast, the NAS Parallel Benchmarks took several years to get six systems on their list.)

"SLALOM netted our fledgling lab its first R&D 100 award and put it on the map. The magazine Supercomputing Review started printing SLALOM performance results as a regular feature."

A number of publications were generated as a result of the SLALOM work:

[RETURN] Back to home page [RETURN] Back to benchmarks page