Carl Friedrich Bolz-Tereick pushed to branch branch/default at PyPy / pypy
Commits: ee32c5a0 by Carl Friedrich Bolz-Tereick at 2021-04-14T12:04:15+02:00 do heap caching for set/getinteriorfield during tracing. this makes the jit.isvirtual inlining decisions less brittle for dictionaries --HG-- branch : jit-heapcache-interiorfields - - - - - ef666b1e by Carl Friedrich Bolz-Tereick at 2021-04-14T12:13:18+02:00 merge default --HG-- branch : jit-heapcache-interiorfields - - - - - d2bf7ba4 by Carl Friedrich Bolz-Tereick at 2021-04-14T12:44:33+02:00 merge default --HG-- branch : jit-heapcache-interiorfields - - - - - 4c3464b8 by Carl Friedrich Bolz-Tereick at 2021-04-14T21:43:37+02:00 ouch, be nan-safe! --HG-- branch : jit-heapcache-interiorfields - - - - - c130b1f2 by Carl Friedrich Bolz-Tereick at 2021-04-15T13:14:01+02:00 fix --HG-- branch : jit-heapcache-interiorfields - - - - - 96d08fba by Carl Friedrich Bolz-Tereick at 2021-04-15T20:42:36+02:00 finally wrote a test, and hopefully fixed the problem really --HG-- branch : jit-heapcache-interiorfields - - - - - ce18a3cd by Carl Friedrich Bolz-Tereick at 2021-04-15T22:05:09+02:00 pfffff, I am not winning this --HG-- branch : jit-heapcache-interiorfields - - - - - 4ef4309f by Carl Friedrich Bolz-Tereick at 2021-04-16T14:59:25+02:00 add some debug output to try to find the problem on 32bit --HG-- branch : jit-heapcache-interiorfields - - - - - 2eb49e81 by Carl Friedrich Bolz-Tereick at 2021-04-16T16:31:26+02:00 test was too strict --HG-- branch : jit-heapcache-interiorfields - - - - - d48fee00 by Carl Friedrich Bolz-Tereick at 2021-04-16T22:08:47+02:00 again what learned. this hopefully works? --HG-- branch : jit-heapcache-interiorfields - - - - - e0b5653a by Carl Friedrich Bolz-Tereick at 2021-04-17T12:30:16+02:00 merge jit-heapcache-interiorfields: add heapcache support for caching get/setinteriorfield. this makes the unrolling decisions based on dicts less brittle. it's not really an optimization on its own, usually, because the dicts that this works on are typically virtual, so they will be removed by optimizeopt anyway. - - - - - 4 changed files: - rpython/jit/metainterp/optimizeopt/test/test_intbound.py - rpython/jit/metainterp/pyjitpl.py - rpython/jit/metainterp/test/test_dict.py - rpython/jit/metainterp/test/test_tracingopts.py View it on GitLab: https://foss.heptapod.net/pypy/pypy/-/compare/87feda40a926f834a7a5c84744269adf6db61368...e0b5653ad3d98b9001f69218220fa20ad8a57aec -- View it on Heptapod: https://foss.heptapod.net/pypy/pypy/-/compare/87feda40a926f834a7a5c84744269adf6db61368...e0b5653ad3d98b9001f69218220fa20ad8a57aec You're receiving this email because of your account on foss.heptapod.net.
_______________________________________________ pypy-commit mailing list pypy-commit@python.org https://mail.python.org/mailman/listinfo/pypy-commit
