Do you need to recompile the kernel because of application requirements or because there's some weird caching bug? It should be fast to load cached kernels. Also, there's a JIT compiler if you can make the necessary modifications in PTX. I haven't used it though.
Native support for multi-threaded garbage collection (it is calling the free function) would be cool, though perhaps not a high priority. regards, Nicholas
_______________________________________________ PyCuda mailing list [email protected] http://tiker.net/mailman/listinfo/pycuda_tiker.net
