[HUF] Improve Huffman encoding speed #2733
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
terrelln
force-pushed
the
huf-cspeed
branch
10 times, most recently
from
8092dde
to
46f2710
Compare
Improve Huffman encoding speed by 20% for gcc and 10% for clang. | Compiler | Benchmark | Config | Dataset | Ratio | Speed MB/s (dev) | Speed MB/s (huf-cspeed) | Speed MB/s (huf-cspeed - dev) | |----------|-------------------|---------|-------------|-------|------------------|-------------------------|-------------------------------| | gcc | compress | level_1 | enwik7 | 2.43 | 253.70 | 258.72 | 2.0% | | gcc | compress | level_1 | silesia | 2.88 | 341.90 | 348.15 | 1.8% | | gcc | compress_literals | level_1 | enwik7 | 1.49 | 761.83 | 912.76 | 19.8% | | gcc | compress_literals | level_1 | silesia | 1.28 | 754.83 | 902.37 | 19.5% | | gcc | compress_literals | level_7 | enwik7 | 1.29 | 502.81 | 552.79 | 9.9% | | gcc | compress_literals | level_7 | silesia | 1.11 | 675.97 | 776.44 | 14.9% | | clang | compress | level_1 | enwik7 | 2.43 | 277.54 | 280.98 | 1.2% | | clang | compress | level_1 | silesia | 2.88 | 369.98 | 375.46 | 1.5% | | clang | compress_literals | level_1 | enwik7 | 1.49 | 828.83 | 918.41 | 10.8% | | clang | compress_literals | level_1 | silesia | 1.28 | 815.81 | 905.41 | 11.0% | | clang | compress_literals | level_7 | enwik7 | 1.29 | 533.13 | 553.30 | 3.8% | | clang | compress_literals | level_7 | silesia | 1.11 | 714.52 | 775.38 | 8.5% |
|
Tests are passing now (except for the 6 that are killed by GitHub actions) |
Cyan4973
reviewed
Jul 30, 2021
| } | ||
| return !bad; | ||
| } | ||
|
|
||
| size_t HUF_compressBound(size_t size) { return HUF_COMPRESSBOUND(size); } | ||
|
|
||
| /** HUF_CStream_t: | ||
| * Huffman uses its own BIT_CStream_t implementation. | ||
| * There are three major differences from BIT_CStream_t: |
Cyan4973
approved these changes
Jul 30, 2021
terrelln
force-pushed
the
huf-cspeed
branch
3 times, most recently
from
e0ef260
to
d8a0797
Compare
* Add a Huffman round trip fuzzer
* Fix two minor bugs in Huffman that aren't exposed in zstd
- Incorrect weight comparison (weights are allowed to be equal to
table log).
- HUF_compress1X_usingCTable_internal() can return compressed
size >= source size, so the assert that `cSize <= 65535` isn't
correct, and it needs to be checked instead.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Improve Huffman encoding speed by 20% for gcc and 10% for clang.
compressbenchmark measures total compression speed.compress_literalsbenchmark compresses the file at the given level. It then extracts the literals from the compressed frame. Then it measures the speed of literal compression on each block of extracted literals.I also added a new Huffman round trip fuzzer in the 2nd commit. It found two minor bugs in Huffman compression that cannot be triggered in zstd. I've run it for a few million iterations and it looks good so far.