Skip to content
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

Jump to bottom

Optimise zlib reader and consolidate sync.Pools #608

Merged
merged 4 commits into from Nov 7, 2022
Merged

Optimise zlib reader and consolidate sync.Pools #608

merged 4 commits into from Nov 7, 2022

Commits on Nov 7, 2022

  1. Use Sync.Pool pointers to optimise memory usage 

    Signed-off-by: Paulo Gomes <pjbgf@linux.com>
    @pjbgf
    pjbgf committed Nov 7, 2022
    Copy the full SHA
    123cdde View commit details
    Browse the repository at this point in the history

  2. Optimise Reference.String() 

    Decreases allocations and bytes per operation by using string builder
with a predefined size.

One additional allocation has been removed by using its own implementation
of Strings(). The reason behind this was due to the fact the calls to
.String() are more recurrent than .Strings() and the performance impact
was worth the code duplication.

Benchmark results:
cpu: 11th Gen Intel(R) Core(TM) i7-11800H @ 2.30GHz
name                        old time/op    new time/op    delta
ReferenceStringSymbolic-16     140ns ± 4%      40ns ± 9%  -71.19%  (p=0.008 n=5+5)
ReferenceStringHash-16         174ns ±14%      85ns ± 4%  -51.13%  (p=0.008 n=5+5)
ReferenceStringInvalid-16     48.9ns ± 2%     1.5ns ± 3%  -96.96%  (p=0.008 n=5+5)

name                        old alloc/op   new alloc/op   delta
ReferenceStringSymbolic-16     88.0B ± 0%     32.0B ± 0%  -63.64%  (p=0.008 n=5+5)
ReferenceStringHash-16          176B ± 0%      144B ± 0%  -18.18%  (p=0.008 n=5+5)
ReferenceStringInvalid-16      0.00B          0.00B          ~     (all equal)

name                        old allocs/op  new allocs/op  delta
ReferenceStringSymbolic-16      4.00 ± 0%      1.00 ± 0%  -75.00%  (p=0.008 n=5+5)
ReferenceStringHash-16          5.00 ± 0%      3.00 ± 0%  -40.00%  (p=0.008 n=5+5)
ReferenceStringInvalid-16       0.00           0.00          ~     (all equal)

Signed-off-by: Paulo Gomes <pjbgf@linux.com>
    @pjbgf
    pjbgf committed Nov 7, 2022
    Copy the full SHA
    ffa7e69 View commit details
    Browse the repository at this point in the history

  3. Optimize zlib reader and consolidate sync.pools 

    Expands on the optimisations from fluxcd#5
and ensures that zlib reader does not need to recreate a deflate
dictionary at every use.

The use of sync pools was consolidated into a new sync utils package.

name       old time/op    new time/op    delta
Parser-16    7.51ms ± 3%    7.71ms ± 6%     ~     (p=0.222 n=5+5)

name       old alloc/op   new alloc/op   delta
Parser-16    4.65MB ± 3%    1.90MB ± 3%  -59.06%  (p=0.008 n=5+5)

name       old allocs/op  new allocs/op  delta
Parser-16     3.48k ± 0%     3.32k ± 0%   -4.57%  (p=0.016 n=5+4)

Signed-off-by: Paulo Gomes <pjbgf@linux.com>
    @pjbgf
    pjbgf committed Nov 7, 2022
    Copy the full SHA
    9490da0 View commit details
    Browse the repository at this point in the history

  4. tests: Replace time.sleep with eventually 

    The previous approach was intermittently flake, leading to different
results based on external results.

The check for goroutines numbers now checks for less or equal, as the
goal of the assertion is to confirm no goroutine is being leaked.

Signed-off-by: Paulo Gomes <pjbgf@linux.com>
    @pjbgf
    pjbgf committed Nov 7, 2022
    Copy the full SHA
    a2c309d View commit details
    Browse the repository at this point in the history