Guillaume Cartier wrote:
I was having a discusion with someone about Gambit and he asked me the following question :
One issue...from the Gambit threading documentation.
"Gambit supports the execution of multiple Scheme threads. These threads are managed entirely by Gambit's runtime and are not related to the host operating system's threads. Gambit's runtime does not currently take advantage of multiprocessors (i.e. at most one thread is running)."
I really do need multiprocessor support as computer graphics gets a pretty linear speedup on multiple CPUs. Do know if Gambit can, like Lispworks, at least have calls into foreign functions running on multiple CPUs?
Does Gambit support that model like Lispworks?
(I haven't used Lispworks.)
Currently there is no builtin support. But you can always run foreign code in separate system threads and communicate with them through pipes, and using macros you could even write a layer atop the builtin FFI to hide the communication handling. I've written cj-posix.scm which is a collection of interfaces to posix procedures and fd->port conversion, anyone interested please ask me for the current version.
In measurements I did some time ago, the cost of a roundtrip through pipes on Linux 2.6 between pthreads of the same process was about 4500 cpu cycles (iirc that was on a Core2 duo in 64bit mode; it was only using read & write, not select inbetween though).
Christian.