You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I suspect this earns the most speedup from floats, where freelist manipulation can be inlined. This might make a single-digit-int freelist more impactful.
Activity
sweeneyde commentedon Jan 25, 2022
GCC --enable-optimizations --with-lto on WSL:
Slower (7):
Faster (27):
Benchmark hidden because not significant (24): chaos, django_template, hexiom, logging_format, logging_simple, pickle, pickle_dict, pidigits, pyflate, python_startup_no_site, raytrace, regex_v8, richards, scimark_lu, sqlalchemy_declarative, sqlalchemy_imperative, sympy_expand, sympy_sum, sympy_str, tornado_http, unpickle, unpickle_list, unpickle_pure_python, xml_etree_generate
Geometric mean: 1.01x faster
sweeneyde commentedon Jan 25, 2022
This attempts to avoid the dispatch dance of
I suspect this earns the most speedup from floats, where freelist manipulation can be inlined. This might make a single-digit-int freelist more impactful.
gh-90667: Add specializations of Py_DECREF when types are known (GH-3…
pythongh-90667: fix incompatible pointer type warnings