We're actually actively working on multithreading and already have a very early implementation that works on Linux and gets pretty good scaling. There's a lot of work left to be done, but the multithreading thing is going to happen sooner rather than later (I could not have said this with any certainty a month ago). That said, Rust has a very well-developed and presumably mature multithreading story, so if you need it now, Rust may be a better bet. If you can hold off for a while and use C++/OpenMP for the critical parallel bits, multithreading is coming to Julia.
On Sat, Jul 19, 2014 at 8:56 AM, Tracy Wadleigh <tracy.wadle...@gmail.com <https://mail.google.com/mail/?view=cm&fs=1&tf=1&to=tracy.wadle...@gmail.com> > wrote: > > FWIW, shared memory parallelism is going to improve in Julia in the > future. > > I've been hoping it might. The issue thread has been open for more than a > year and a half and silent the last six months, so I've been getting antsy. > > Maybe I will just continue to bide my time by sticking with C++/OpenMP for > certain critical bits until Julia can catch up. (I'd help with this task if > I thought I could contribute in any net-positive way.) > > Meanwhile, Rust does look like a pretty legit contender to unseat C++. It > includes a bunch of features I love from my other favorite language > (Haskell), like algebraic data types, pattern matching, and type classes > (although extending type classes to higher kinded types is a currently > missing must-have), along with a fancy memory model worthy of the > multi-core age in which we live. I only wish they might have chosen to > borrow a little more heavily from my most favorite feature of Haskell: it's > beautiful, beautiful syntax. >
