Jitviewer and performance tuning
AJ R <[email protected]> Mon, 8 Aug 2022 18:06:58 +0200
| Newsgroups | gmane.comp.python.pypy |
|---|---|
| Message-ID | <CAL+nYz9gPXvXcXU_J22+gJFE_=dFWgU0e7oPd899RKoi1kCzyQ@mail.gmail.com> |
Hi, I converted a long running program (~52k LOC) from python 2.7 to python 3.8. I run my tests on Window 10 (64) against pypy 3.8 (7.3.7) and cpython 3.8. Now I'm profiling the code with cProfile + gprof2dot. I noticed that increasing PYPY_GC_NURSERY improved the program's performance, but it's still only marginally faster than under Cpython. I set PYPYLOG to generate a Jitviewer-compatible file (~ 17Mb), but it seems that this tool is no longer maintained. Is that correct? What can I do to get more insight into what the JIT is doing? And what GC vars should generally be used? Thanks in advance for your help. Best wishes, Albert-Jan _______________________________________________ pypy-dev mailing list -- [email protected] To unsubscribe send an email to [email protected] https://mail.python.org/mailman3/lists/pypy-dev.python.org/ Member address: [email protected]