Frequently Asked Questions & Known Bugs

Currently, we only support Linux x86-64bit architecture. We have not recently tested on Mac OS X Intel-64bit architecture, but it will probably not work. For instance, the last stable version of Debian (or Ubuntu) Linux x86-64bit should be fine.

Are you using unsupported functions ? Some functions are not yet available, they should not be used. See download page.

Else, please send us a bug report.

Check if your program contains non-allocating threads, if it does, those may prevent other threads to run when a memory needs to be collected.

The heap (where memory is allocated) may not be big enough. See the README file to know how to parameterize the heap size.

Thread.kill is not available in INRIA's official OCaml distribution with ocamlopt (Thread.kill only works with VM threads of ocamlc/ocamlrun).

How many threads do you run in parallel ? Do you have a lot of memory access ? Parallel threads reduce the per-thread memory bandwidth, you can try to make fewer parallel memory access.

How much is your program allocating ? If a lot, our GC is not as performant as INRIA's.

If your program reaches a dead-lock while it really doesn't with official version, please send a bug report.

With n CPU-cores, you can expect a speed-up factor of n if the memory accesses do not kill your program performances. Sometimes you can have an even greater speed-up if you manage to reduce thread-context-switches or page faults.

With intensive memory-access, the speed-up given by the numerous cores will be dropped by the high cost of parallel or concurrent memory accesses.

If you can write your program without mutexes, then you should do it. Mutexes are a very high cost mechanism, which can sometimes become neglectful.

Take advantage of the growing number of CPU-cores available in our PCs.

Run computation faster.