René's URL Explorer Experiment

Title: gh-94603: micro optimize list.pop by eendebakpt · Pull Request #94604 · python/cpython · GitHub

Open Graph Title: gh-94603: micro optimize list.pop by eendebakpt · Pull Request #94604 · python/cpython

X Title: gh-94603: micro optimize list.pop by eendebakpt · Pull Request #94604 · python/cpython

Description: This PR optmizes list.pop Benchmark results x=[1,2]; x.pop(0): Mean +- std dev: [m2] 81.8 ns +- 2.5 ns -> [pr_special] 68.9 ns +- 2.0 ns: 1.19x faster x=[1,2]; x.pop(1): Mean +- std dev: [m2] 65.4 ns +- 2.1 ns -> [pr_special] 64.4 ns +- 0.6 ns: 1.02x faster x=[]; x.insert(0,1); x.pop(0): Mean +- std dev: [m2] 111 ns +- 3 ns -> [pr_special] 80.2 ns +- 1.1 ns: 1.38x faster list(0): insert(0, None) + pop(0): Mean +- std dev: [m2] 78.4 ns +- 0.8 ns -> [pr_special] 62.1 ns +- 1.3 ns: 1.26x faster list(1): insert(0, None) + pop(1): Mean +- std dev: [m2] 59.7 ns +- 0.4 ns -> [pr_special] 57.7 ns +- 0.9 ns: 1.04x faster list(2): insert(0, None) + pop(1): Mean +- std dev: [m2] 66.9 ns +- 0.6 ns -> [pr_special] 55.5 ns +- 1.8 ns: 1.20x faster list(10): insert(0, None) + pop(1): Mean +- std dev: [m2] 71.9 ns +- 0.6 ns -> [pr_special] 62.4 ns +- 2.8 ns: 1.15x faster list(100): insert(0, None) + pop(1): Mean +- std dev: [m2] 108 ns +- 4 ns -> [pr_special] 93.5 ns +- 3.3 ns: 1.16x faster list(10000): insert(0, None) + pop(1): Mean +- std dev: [m2] 5.41 us +- 0.02 us -> [pr_special] 5.37 us +- 0.02 us: 1.01x faster Benchmark hidden because not significant (2): x=[1]; x.pop(0), sortedcontainer Geometric mean: 1.12x faster Benchmark code import pyperf import sortedcontainers runner = pyperf.Runner() runner.timeit(name=f"x=[1]; x.pop(0)",stmt='x=[1]; x.pop(0)') runner.timeit(name=f"x=[1,2]; x.pop(0)", stmt='x=[1,2]; x.pop(0)') runner.timeit(name=f"x=[1,2]; x.pop(1)",stmt='x=[1,2]; x.pop(1)') runner.timeit(name=f"x=[]; x.insert(0,1); x.pop(0)",stmt='x=[]; x.insert(0,1); x.pop(0)') runner.timeit(name=f"list(0): insert(0, None) + pop(0)", stmt="lst.insert(0, None); lst.pop(0)", setup='lst=[]') for n in [ 1, 2, 10, 100, 10_000]: runner.timeit(name=f"list({n}): insert(0, None) + pop(1)", stmt="lst.insert(0, None); lst.pop(1)", setup=f'import random; lst=[random.random() for ii in range({n})]') setup ="import random; import sortedcontainers; sc=sortedcontainers.SortedList(); x=[random.random() for n in range(1_00_000)]" runner.timeit(name=f"sortedcontainer", stmt="[sc.add(v) for v in x]", setup=setup) Issue: gh-94603

Open Graph Description: This PR optmizes list.pop Benchmark results x=[1,2]; x.pop(0): Mean +- std dev: [m2] 81.8 ns +- 2.5 ns -> [pr_special] 68.9 ns +- 2.0 ns: 1.19x faster x=[1,2]; x.pop(1): Mean +- std dev: [m2] 65...

X Description: This PR optmizes list.pop Benchmark results x=[1,2]; x.pop(0): Mean +- std dev: [m2] 81.8 ns +- 2.5 ns -> [pr_special] 68.9 ns +- 2.0 ns: 1.19x faster x=[1,2]; x.pop(1): Mean +- std dev: [m2...

Opengraph URL: https://github.com/python/cpython/pull/94604

X: @github

direct link

Domain: github.com

Links:

Viewport: width=device-width