* Laslo Hunhold <dev_AT_frign.de> [2021-05-04 09:17:01 +0200]:
>On Sun, 02 May 2021 18:17:45 -0400
>"Greg Reagle" <greagle_AT_fastmail.fm> wrote:
>
>Dear Greg,
>
>> Do you have any other suggestions for alternatives to C?
>
>this question is too general. For academic purposes (HPC, data
>analysis, numerical mathematics, statistics, etc.) I can recommend
>Julia.
If you want to do simple tasks on your workstation like solve some ODEs,
Julia is a viable alternative to MATLAB/Python because of its relative
speed and ease of use.
However, I would *never* consider Julia a viable alternative to C/FORTRAN
tasks, including numerical simulations and massively parallel deployment
on HPC systems. I've worked with both Julia and C, and strongly advise
against Julia for those purposes. While promising on paper, the reality
is that the language is immature which creates issues with code
compability between versions. One example is that they flip-flop between
variable scope in global space, effectively breaking most scripts without
warning. Also, the julia+python+blas dependencies of installed packages
and computational overhead quickly become very significant. Furthermore,
the garbage collection is poor and leads to orders-of-magnitude increase
in memory footprint over days of running an iterative simulation.
--
Anders Damsgaard
https://adamsgaard.dk
gopher://adamsgaard.dk
Received on Wed May 05 2021 - 15:48:16 CEST