Hi Marco, Characterizing "a complex simulation" to "compute pretty fast" is not so simple. You need to consider what sort of models you will be solving (vis. the level of parallelization that is achievable), the size and discretization, and other factors. I would suggest to discuss this with the people who are developing the simulations and find out what hardware they require.
As far as Julia performance: well-written Julia code can typically come within a low constant factor of C++ performance (generally < 2x). Best, Isaiah On Thu, Aug 13, 2015 at 6:31 AM, Marco Pleines <[email protected]> wrote: > Hello everybody, > > I'm participating in the process of establishing a math simulation lab on > my college. Right now I'm gathering information to determine a reasonable > computer setup to use with Julia. > > My question is pretty simple: > > Is it possible to compute the mentioned simulations with a cluster of 3 > computers (featuring Intel Skylake and nVidia's GTX 980ti)? > The goal is to compute pretty fast (close to real-time) complex > simulations. > > > So far we came across Julia and think that we should give it a try. > > >
