These tests were developed for vector machines and performance numbers were obtained to present at the ParkBench meeting in Southampton in September 1997, even though the two machines these tests were run on are not vector machines.

Many of these tests were looped for timing purposes. Since the secondary cache of the Origin is much larger than the secondary cache of the T3E (4 MB versus 96 KB), the reader should be aware that the looping of the performance tests that fit into Origin's cache but not into the T3E's cache will greatly enhance the performance number obtained on the Origin relative to the performance number obtained on the T3E.