[Gllug] Programmer Wanted (scripter and compiler)

John Hearns hearnsj at googlemail.com
Wed May 6 09:30:29 UTC 2009


2009/5/6 Nix <nix at esperi.org.uk>:
>>
> I like. (yes, I could use a less brute-force technique, but it's useful
> to know that the crude approaches *work* if you have insane amounts of
> machine time :))) )


When I was but a cub researcher, I worked on cancer therapy
optimisation for conformal radiotherapy - ie. shaping therapy beams to
the tumour shape.
At the time I was interested in silulated annealing, in order to
optimise the solution.
I had a Sparc 1+ with all of 2 gigs of RAM (which was an upgraqde on
the standard 1gig). Even a simplified 2D problem would swap horribly,
and I had to start a run and come back the next moring to see if it
worked.
A comment once made to me was that for medical work the results have
to come back in an horu or two - during the duration of a clinic. The
same goes in many other fields - 'time to solution' is important.

So I absolutely agree with what you say - increasing CPU power, and
harnessing parallel facilities, makes new types of work feasible.
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug




More information about the GLLUG mailing list