Author: Remi Meier <[email protected]> Branch: stmgc-c8-gcc Changeset: r79612:f7db687c6f7b Date: 2015-09-14 10:48 +0200 http://bitbucket.org/pypy/pypy/changeset/f7db687c6f7b/
Log: update TODO diff --git a/TODO b/TODO --- a/TODO +++ b/TODO @@ -1,15 +1,3 @@ ------------------------------------------------------------- - -## maybe resolved: -also a possible cause for the thing below this one: - pypy-stm frees 530936 loop tokens - pypy26 frees 446 loop tokens -meaning, loops get freed and probably must be retraced all -the time... see XXXXXX in memmgr.py - -one problem was wrong counting, so now its 1551 vs. 446. Still -probably too much. - ------------------------------------------------------------ annotating and rtyping pixie with --jit=off and 1 thread (no @@ -17,6 +5,10 @@ STM: 148s pypy2.6: 91.3s +full mapdict and methodcache bring STM down to 130s, which +is a reasonable 40% overhead. --> think about fixing mapdict +and methodcache with STM + This difference may also explain a similar with-jit result: STM: 96 pypy2.6: 64.5s @@ -25,8 +17,8 @@ pypy2.6: 64s (still using only 1 thread) TODOs: -1) stmdict/stmset need strategies -2) methodcache missing or generally no-jit performance is bad here +1) stmdict/stmset need strategies (only ~2% perf here) +2) methodcache & mapdict for generally better no-jit performance 3) theoretically, more threads shouldn't take even longer ------------------------------------------------------------ _______________________________________________ pypy-commit mailing list [email protected] https://mail.python.org/mailman/listinfo/pypy-commit
