* Peter Verswyvelen <bugfact@gmail.com> [2009-03-08 12:00:23+0100]
On Sat, Mar 7, 2009 at 6:49 PM, Roman Cheplyaka <roma@ro-che.info> wrote:
* Peter Verswyvelen <bugfact@gmail.com> [2009-03-07 18:34:10+0100]
On Fri, Mar 6, 2009 at 10:39 PM, Roman Cheplyaka <roma@ro-che.info> wrote:
Is it a performance issue?
Yes. As Roman Leshchinskiy explained it's because parallel arrays are not fused properly yet. But it's quite trivial to modify Hpysics to use ordinary lists or arrays, and I think it'll give reasonable performance.
Personally I would not mind too much about performance initially; I'm very confident that lots of optimizations can be done afterwards - maybe even letting the GPU handle certain tasks.
I wouldn't mind too if it was just 'inefficient'; however, it's practically unusable. <<Result was disappointing -- simple 2-step simulation with 10 bodies took 9 seconds and almost 1Gb of memory to complete.>> http://physics-dph.blogspot.com/2008/08/status-report-week-11-12.html -- Roman I. Cheplyaka :: http://ro-che.info/ "Don't let school get in the way of your education." - Mark Twain