In article <33147 at castle.ed.ac.uk> sss at castle.ed.ac.uk (S S Sturrock) writes:
>Just one thing for the computer scientists reading this, for the most part
>Teraflops (or Gigaflops at the moment) are really not very useful for the
>day to day biology task, ie sequence analysis. Sure for modelling etc they
>become quite handy but integer arithmetic is where it is still at.
The reason that people measure things in flops is that most of the
applications driving the development of massively parallel machines
are in the scientific/engineering arena, where flops are the goal.
For my own work, flops are also irrelevant. Many of the processors
being used favor vectorizable floating point operations, including the
CM-5 and the new Meiko machine. Last time I checked, these machines
can be ordered without vector units, but few people do. For a machine
like the Intel Paragon or nCUBE-2, the correspondence between flops
and mips is pretty close anyway.
>and although the CM-200 claims to have 8 Gigaflops performance (I am sick
>and tired of people claiming it is the most powerful supercomputer in the
>UK)
We recently turned ours off and boxed it up. Too out of date, and too
quirky (SIMD, after all!).
Think in terms of what you could do with a machine with
- 128 Gbytes RAM
- 1 Terabyte disk
- 300,000 MIPS
Such a machine is likely to appear in the next two years.
Kevin McCurley
Sandia National Laboratories
