Christian Tismer wrote: > The Q1 goals are relatively doable without doubt. The current > achievements speedwise remind me of the anxient Python2C project. > It showed the typical acceleration by a factor of around 2, which > is what you can expect when eliminating the interpreter loop.
A bit less than that, but this sounds about right. Last I tried (somewhere before the 0.10 release), Cython gave you a total of about 10-30% for (most of) pybench, some things (like loops) being really more in the order of a factor of 2 to 6. I'd expect it to be another bit more in 0.12. >> "Eliminate the GIL" is not hard by itself [...] > As I remember that patch, the overhead was around 40%, mostly because > of reference counting. I guess nobody actually goes this path, > because it is such a waste, compared to multiple processes, and doing > it "right" (where I'm referring to some Java achievements I heard of) > is pretty much of a total re-write of Python. > > I'm pretty much wondering if the latter makes sense, given the > existence of PyPy. ... and Cython. The fact that Cython uses CPython's C-API doesn't mean that it's not in the same order as a Python implementation. We just happily reuse what's there already - and we happily use it to interface with what's there already. Stefan _______________________________________________ [email protected] http://codespeak.net/mailman/listinfo/pypy-dev
