Hi, On Fri, Mar 8, 2013 at 8:08 PM, GW <gw.2...@tnode.com> wrote: > I am wondering why is there a proposal to provide only a compatible working > NumPy CPU implementation for PyPy? Aren't NumPy operations more suited for > GPGPU (eg. OpenCL or CUDA) and supporting them would result in a greater > performance boost than just CPU?
PyPy doesn't include a "JIT for NumPy": it includes a JIT for the whole Python language, which can also work for Python programs that use NumPy. As far as I know you cannot run with reasonable performance a complete, unrestricted Python program on a GPGPU. Whether PyPy's NumPy will, over time, be extended to also support GPGPU operations --- and whether this would then benefit from changes to the JIT subcomponent itself --- only time and volunteers can say. A bientôt, Armin. _______________________________________________ pypy-dev mailing list pypy-dev@python.org http://mail.python.org/mailman/listinfo/pypy-dev
