gh-109868: Skip deepcopy memo check for empty memo

[eendebakpt]

In the copy.deepcopy method we can skip the initial memo check if the memo was just created.

Benchmark:

from copy import deepcopy
obj={'simple': 'dict'}
%timeit deepcopy(obj)

main: 1.91 µs ± 1.38 ns per loop (mean ± std. dev. of 7 runs, 1,000,000 loops each)
pr: 1.82 µs ± 8.18 ns per loop (mean ± std. dev. of 7 runs, 1,000,000 loops each)

• Issue: Skip deepcopy memo check when memo is empty #109868

[serhiy-storchaka]

The difference is equal to one standard deviation. I am not sure that it is not random.

[eendebakpt]

@serhiy-storchaka Here are some more detailed measurements:

import pyperf

setup = """
from copy import deepcopy
d1 = 10
d2 = [1, 2]
d3 = {'hello', 'there'}
d4 = ([1], [2])
"""

runner = pyperf.Runner()
for ii in [1, 2, 3, 4]:
	runner.timeit(name=f"d{ii}", stmt=f"deepcopy(d{ii})", setup=setup)

Results (with --rigorous option):

d1: Mean +- std dev: [main] 275 ns +- 5 ns -> [pr] 226 ns +- 6 ns: 1.22x faster
d2: Mean +- std dev: [main] 1.73 us +- 0.02 us -> [pr] 1.69 us +- 0.04 us: 1.02x faster
d3: Mean +- std dev: [main] 5.98 us +- 0.09 us -> [pr] 5.92 us +- 0.09 us: 1.01x faster
d4: Mean +- std dev: [main] 3.83 us +- 0.07 us -> [pr] 3.81 us +- 0.05 us: 1.01x faster

Geometric mean: 1.06x faster

The code change itself is very simple. It avoids a dict lookup, assignment to a local variable y and a comparison, which provides a small speedup. For a deepcopy of a larger object the performance improvement will indeed be below the noise.

The performance gain is small, but so is the change and I do not see any downsides to the change.

[serhiy-storchaka]

Well, the benefit is tiny, but the cost is virtually zero. I think nothing bad will happen if we accept it.

Thank you for your contribution @eendebakpt.

[bedevere-bot]

⚠️⚠️⚠️ Buildbot failure ⚠️⚠️⚠️

Hi! The buildbot AMD64 Debian root 3.x has failed when building commit 05079d9.

What do you need to do:

1. Don't panic.
2. Check the buildbot page in the devguide if you don't know what the buildbots are or how they work.
3. Go to the page of the buildbot that failed (https://buildbot.python.org/all/#builders/345/builds/5956) and take a look at the build logs.
4. Check if the failure is related to this commit (05079d9) or if it is a false positive.
5. If the failure is related to this commit, please, reflect that on the issue and make a new Pull Request with a fix.

You can take a look at the buildbot page here:

https://buildbot.python.org/all/#builders/345/builds/5956

Failed tests:

• test_signal

Summary of the results of the build (if available):

==

Click to see traceback logs

Traceback (most recent call last):
  File "/root/buildarea/3.x.angelico-debian-amd64/build/Lib/test/test_signal.py", line 1287, in test_stress_delivery_dependent
    self.assertEqual(len(sigs), N, "Some signals were lost")
AssertionError: 5220 != 10000 : Some signals were lost