> Another consideration is stability. Suppose you've got some real
>number-crunching to do, and your algorithm is going to take eight hours
>to complete. Do you want to trust it to an OS that's going to
>crash half-way through (or at the last second)? Closed-source
>systems are too unpredictable for such applications, which is
>why Linux and its relatives are becoming increasingly popular
>among physicists and the like. I'd be interested to know if there is a
>version of Mathematica or Maple (or another good CAS) for Linux.
Closed source systems are not inherently unpredictable. Solaris and Irix,
in my experience, do not crash, and there's a reason why Sun has a very
good position in the Unix server business. (Except when using Netscape.
Everything crashes using Netscape. The universal constant of desktop
computing.) I'm not a systems programmer by any stretch of the
imagination, so it does me no good to have the Linux source, as opposed to
using Irix. I'm sure either would work, but the truth is there's probably
no good reason for NT to be less stable than Linux at this early point in
both OS's development, except shoddy engineering in Redmond.
If I were going to run an eight hour algorithm, though, I doubt I'd use
anything that'd run NT. Does anyone actually use NT for number crunching
of that sort?
-Nat
This archive was generated by hypermail 2b29 : Wed Apr 27 2005 - 03:30:03 EDT