8000 ENH: Provides a deep benchmark for universal functions by seiko2plus · Pull Request #15992 · numpy/numpy · GitHub
[go: up one dir, main page]
Skip to content

ENH: Provides a deep benchmark for universal functions #15992

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

ENH: Provides a deep benchmark for universal functions #15992

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
69 changes: 69 additions & 0 deletions benchmarks/benchmarks/bench_deep.py
View file
Open in desktop
8000
Original file line number Diff line number Diff line change
@@ -0,0 +1,69 @@
"""
Provides a deep benchmark for ufunc(universal functions).

Usage:
deep.{name}_{input types}
Example:
--bench-compare master "deep.square_(h|H)"
"""
import numpy as np
from textwrap import dedent
from functools import lru_cache

@lru_cache(maxsize=1024)
def rand(size, dtype, prevent_overlap=0):
if dtype == '?':
return np.random.randint(0, 1, size=size, dtype=dtype)
elif dtype in 'bBhHiIlLqQ':
return np.random.randint(1, 127, size=size, dtype=dtype)
else:
return np.array(np.random.rand(size), dtype=dtype)

class _Benchmark:
number = 10
repeat = (100, 100, 0)

def setup(self, size, *strides):
np.seterr(all='ignore')
self.fn = getattr(np, self.ufunc)
self.args = [
rand(size * strides[c], t, c)[::strides[c]]
for c, t in enumerate(self.arg_types)
]

def time_run(self, *args):
self.fn(*self.args)

sizes = [256, 4096, 16384]
strides = [1, 2, 5]
klasses = []

for name in dir(np):
ufunc = getattr(np, name)
if not isinstance(ufunc, np.ufunc):
continue
param_names = ["size"]
param_names += ["stride_in%d" % i for i in range(ufunc.nin)]
param_names += ["stride_out%d" % i for i in range(ufunc.nout)]
param_names = repr(param_names)
params = [sizes]
params += [strides] * (ufunc.nin + ufunc.nout)
params = repr(params)

for tsym in ufunc.types:
tsym = tsym.split('->')
if len(tsym) != 2:
continue
klasses.append(dedent("""
class {name}_{dtypes}(_Bench 51CA mark):
params={params}
param_names={param_names}
ufunc="{name}"
arg_types="{arg_types}"
""".format(
name=name, dtypes=('_'.join(tsym)).replace('?', 'bool'),
params=params, param_names=param_names,
arg_types=tsym[0]+tsym[1]
)))

exec('\n'.join(klasses))
0