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.

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

zstd decode: Use compound decSymbol #144

Merged
merged 3 commits into from
Aug 3, 2019
Merged

Conversation

klauspost
Copy link
Owner

@klauspost klauspost commented Jul 28, 2019

The compiler is not generating very nice code when decSymbol is separate values.

Use a compound value instead of a struct.

name                                     old time/op    new time/op    delta
Decoder_DecodeAll/kppkn.gtb.zst-12          604µs ± 1%     600µs ± 0%   -0.76%        (p=0.000 n=10+10)
Decoder_DecodeAll/geo.protodata.zst-12      136µs ± 1%     135µs ± 0%   -0.83%         (p=0.000 n=10+9)
Decoder_DecodeAll/plrabn12.txt.zst-12      1.94ms ± 1%    1.92ms ± 1%   -1.19%        (p=0.000 n=10+10)
Decoder_DecodeAll/lcet10.txt.zst-12        1.45ms ± 1%    1.43ms ± 0%   -1.69%         (p=0.000 n=10+8)
Decoder_DecodeAll/asyoulik.txt.zst-12       506µs ± 1%     497µs ± 1%   -1.78%         (p=0.000 n=9+10)
Decoder_DecodeAll/alice29.txt.zst-12        647µs ± 1%     641µs ± 0%   -0.85%         (p=0.000 n=10+9)
Decoder_DecodeAll/html_x_4.zst-12           263µs ± 1%     305µs ± 1%  +16.17%        (p=0.000 n=10+10)
Decoder_DecodeAll/paper-100k.pdf.zst-12    23.8µs ± 1%    23.8µs ± 1%     ~           (p=0.171 n=10+10)
Decoder_DecodeAll/fireworks.jpeg.zst-12    9.57µs ± 1%    9.56µs ± 0%     ~            (p=0.590 n=10+9)
Decoder_DecodeAll/urls.10K.zst-12          1.65ms ± 0%    1.64ms ± 1%     ~            (p=0.403 n=7+10)
Decoder_DecodeAll/html.zst-12               156µs ± 1%     154µs ± 1%   -0.85%        (p=0.012 n=10+10)

name                                     old speed      new speed      delta
Decoder_DecodeAll/kppkn.gtb.zst-12        305MB/s ± 1%   307MB/s ± 0%   +0.77%        (p=0.000 n=10+10)
Decoder_DecodeAll/geo.protodata.zst-12    869MB/s ± 1%   876MB/s ± 0%   +0.84%         (p=0.000 n=10+9)
Decoder_DecodeAll/plrabn12.txt.zst-12     249MB/s ± 1%   252MB/s ± 1%   +1.21%        (p=0.000 n=10+10)
Decoder_DecodeAll/lcet10.txt.zst-12       294MB/s ± 1%   299MB/s ± 0%   +1.72%         (p=0.000 n=10+8)
Decoder_DecodeAll/asyoulik.txt.zst-12     247MB/s ± 1%   252MB/s ± 1%   +1.81%         (p=0.000 n=9+10)
Decoder_DecodeAll/alice29.txt.zst-12      235MB/s ± 1%   237MB/s ± 0%   +0.86%         (p=0.000 n=10+9)
Decoder_DecodeAll/html_x_4.zst-12        1.56GB/s ± 1%  1.34GB/s ± 1%  -13.92%        (p=0.000 n=10+10)
Decoder_DecodeAll/paper-100k.pdf.zst-12  4.30GB/s ± 1%  4.31GB/s ± 1%     ~           (p=0.172 n=10+10)
Decoder_DecodeAll/fireworks.jpeg.zst-12  12.9GB/s ± 1%  12.9GB/s ± 0%     ~            (p=0.646 n=10+9)
Decoder_DecodeAll/urls.10K.zst-12         425MB/s ± 0%   427MB/s ± 1%     ~            (p=0.403 n=7+10)
Decoder_DecodeAll/html.zst-12             658MB/s ± 1%   663MB/s ± 1%   +0.86%        (p=0.008 n=10+10)

The compiler is not generating very nice code when decSymbol is separate values.

Use a compound value instead of a struct.

```
name                                     old time/op    new time/op    delta
Decoder_DecodeAll/kppkn.gtb.zst-12          610µs ± 1%     599µs ± 1%  -1.84%        (p=0.000 n=10+10)
Decoder_DecodeAll/geo.protodata.zst-12      138µs ± 2%     136µs ± 0%  -1.72%         (p=0.000 n=10+9)
Decoder_DecodeAll/plrabn12.txt.zst-12      1.95ms ± 1%    1.90ms ± 0%  -2.19%         (p=0.000 n=10+7)
Decoder_DecodeAll/lcet10.txt.zst-12        1.46ms ± 2%    1.42ms ± 1%  -2.48%        (p=0.000 n=10+10)
Decoder_DecodeAll/asyoulik.txt.zst-12       506µs ± 1%     497µs ± 1%  -1.83%         (p=0.000 n=10+9)
Decoder_DecodeAll/alice29.txt.zst-12        655µs ± 1%     636µs ± 1%  -2.97%        (p=0.000 n=10+10)
Decoder_DecodeAll/html_x_4.zst-12           267µs ± 1%     261µs ± 1%  -2.43%         (p=0.000 n=10+9)
Decoder_DecodeAll/paper-100k.pdf.zst-12    25.0µs ± 1%    24.3µs ± 1%  -2.61%        (p=0.000 n=10+10)
Decoder_DecodeAll/fireworks.jpeg.zst-12    9.65µs ± 1%    9.61µs ± 0%    ~            (p=0.250 n=10+9)
Decoder_DecodeAll/urls.10K.zst-12          1.67ms ± 1%    1.63ms ± 2%  -2.29%        (p=0.000 n=10+10)
Decoder_DecodeAll/html.zst-12               156µs ± 1%     155µs ± 1%  -0.85%        (p=0.014 n=10+10)

name                                     old speed      new speed      delta
Decoder_DecodeAll/kppkn.gtb.zst-12        302MB/s ± 1%   308MB/s ± 1%  +1.88%        (p=0.000 n=10+10)
Decoder_DecodeAll/geo.protodata.zst-12    860MB/s ± 2%   875MB/s ± 0%  +1.75%         (p=0.000 n=10+9)
Decoder_DecodeAll/plrabn12.txt.zst-12     248MB/s ± 1%   253MB/s ± 0%  +2.24%         (p=0.000 n=10+7)
Decoder_DecodeAll/lcet10.txt.zst-12       293MB/s ± 2%   300MB/s ± 1%  +2.54%        (p=0.000 n=10+10)
Decoder_DecodeAll/asyoulik.txt.zst-12     247MB/s ± 1%   252MB/s ± 1%  +1.86%         (p=0.000 n=10+9)
Decoder_DecodeAll/alice29.txt.zst-12      232MB/s ± 1%   239MB/s ± 1%  +3.06%        (p=0.000 n=10+10)
Decoder_DecodeAll/html_x_4.zst-12        1.53GB/s ± 1%  1.57GB/s ± 1%  +2.49%         (p=0.000 n=10+9)
Decoder_DecodeAll/paper-100k.pdf.zst-12  4.10GB/s ± 1%  4.21GB/s ± 1%  +2.68%        (p=0.000 n=10+10)
Decoder_DecodeAll/fireworks.jpeg.zst-12  12.8GB/s ± 1%  12.8GB/s ± 0%    ~            (p=0.286 n=10+9)
Decoder_DecodeAll/urls.10K.zst-12         420MB/s ± 1%   430MB/s ± 2%  +2.35%        (p=0.000 n=10+10)
Decoder_DecodeAll/html.zst-12             655MB/s ± 1%   661MB/s ± 1%  +0.86%        (p=0.015 n=10+10)
```
@klauspost klauspost merged commit 0e54620 into master Aug 3, 2019
@klauspost klauspost deleted the zstd-compund-decSymbol branch August 3, 2019 08:50
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant