On Mon, Apr 12, 2010 at 11:00 AM, René Dudfield <[email protected]> wrote: > ah, indeed. oops. > > It looks like (jager+tracing) is now similar in speed to tracing (still a > bit slower), whereas (jager+tracing) used to be closer to the interpreter > speed. Whereas jager is only slightly faster than the base interpreter. > > cu. >
Please note that those benchmarks are extremely JIT-friendly in general. They have very tight loops doing one operation or mostly something like that. _______________________________________________ [email protected] http://codespeak.net/mailman/listinfo/pypy-dev
