rapidgzip
TurboBench
rapidgzip | TurboBench | |
---|---|---|
14 | 10 | |
320 | 312 | |
- | - | |
9.5 | 8.9 | |
10 days ago | 9 months ago | |
C++ | C | |
Apache License 2.0 | - |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
rapidgzip
- Show HN: Rapidgzip – Parallel Gzip Decompressing with 10 GB/S
-
Ebiggers/libdeflate: Heavily optimized DEFLATE/zlib/gzip library
I also did benchmarks with zlib and libarchivemount via their library interface here [0]. It has been a while that I have run them, so I forgot. Unfortunately, I did not add libdeflate.
[0] https://github.com/mxmlnkn/rapidgzip/blob/master/src/benchma...
-
Rapidgzip – Parallel Decompression and Seeking in Gzip (Knespel, Brunst – 2023) [pdf]
Hi, author here.
You are right in the index being the easy-mode. Over the years there have been lots of implementations trying to add an index like that to the gzip metadata itself or as a sidecar file, with bgzip probably being the most known one. None of them really did stick, hence the necessity for some generic multi-threaded decompressor. A probably incomplete list of such implementations can be found in this issue: https://github.com/mxmlnkn/rapidgzip/issues/8
The index makes it so easy that I can simply delegate decompression to zlib. And since paper publication I've actually improved upon this by delegating to ISA-l / igzip instead, which is twice as fast. This is already in the 0.8.0 release.
As derived from table 1, the false positive rate is 1 Tbit / 202 = 5 Gbit or 625 MB for deflate blocks with dynamic Huffman code. For non-compressed blocks, the false positive rate is roughly one per 500 KB, however non-compressed blocks can basically be memcpied or skipped over and then the next deflate header can be checked without much latency. On the other hand, for dynamic blocks, the whole block needs to be decompressed first to find the next one. So the much higher false positive rate for non-compressed blocks doesn't introduce that much overhead.
I have some profiling built into rapidgzip, which is printed with -v, e.g., rapidgzip -v -d -o /dev/null 20xsilesia.tar.gz :
Time spent in block finder : 0.227751 s
- Intel QuickAssist Technology Zstandard Plugin for Zstandard
- Tool and Library for Parallel Gzip Decompression and Random Access
-
Pigz: Parallel gzip for modern multi-processor, multi-core machines
I have not only implemented parallel decompression but also random access to offsets in the stream with https://github.com/mxmlnkn/pragzip I did some benchmarks on some really beefy machines with 128 cores and was able to reach almost 20 GB/s decompression bandwidth. The single-core decoder has lots of potential for optimization because I had to write it from scratch, though.
-
Parquet: More than just “Turbo CSV”
Decompression of arbitrary gzip files can be parallelized with pragzip: https://github.com/mxmlnkn/pragzip
-
The Cost of Exception Handling
At the very least you are duplicating logic without the exception. The check for eof has to be done implicitly anyway inside read because it has to fill the bit buffer with data from the byte buffer or the byte buffer with data from the file. And if both fail, then we already know the result of eof, so no need to duplicate checking for eof in the outer read calling loop.
Here is the full commit with ad-hoc benchmark results in the commit message:
https://github.com/mxmlnkn/pragzip/commit/0b1af498377838c30f...
and here the benchmarks I ran at that time:
https://github.com/mxmlnkn/pragzip/blob/0b1af498377838c30fea...
As you can see, it's part of my random-seekable multi-threaded gzip and bzip2 parallel decompression libraries.
What you can also see in the commit message is that it wasn't a 50% time reduction but a 50% bandwidth increase, which would translate to a 30% time reduction. It seems I remembered that partly wrong. But it still was a significant optimization for me.
- How Much Faster Is Making a Tar Archive Without Gzip?
- Show HN: Thread-Parallel Decompression and Random Access to Gzip Files (Pragzip)
TurboBench
-
Zstd Content-Encoding planned to ship with Chrome 123
I'm still unconvinced about this addition. And I don't even dislike Zstandard.
The main motivation seems to be that while Zstandard is worse than Brotli at the highest level, it's substantially faster than Brotli when data has to be compressed on the fly with a limited computation budget. That might be true, but I'm yet to see any concrete or even anecdotal evidence even in the issue tracker [1] while there exist some benchmarks where both Zstandard and Brotli are fast enough for the web usage even at lower levels [2].
According to their FAQ [3] Meta and Akamai have successfully used Zstandard in their internal network, but my gut feeling is that they never actually tried to optimize Brotli instead. In fact, Meta employs the main author of Zstandard so it would have been easier to tune Zstandard instead of Brotli. While Brotli has some fundamental difference from Zstandard (in particular Brotli doesn't use arithmetic-equivalent coding), no one has concretely demonstrated that difference would prevent Brotli from being fast enough for dynamic contents in my opinion.
[1] https://issues.chromium.org/issues/40196713
[2] https://github.com/powturbo/TurboBench/issues/43
[3] https://docs.google.com/document/d/14dbzMpsYPfkefAJos124uPrl...
- TurboBench: Dynamic/Static web content compression benchmark
-
Ebiggers/libdeflate: Heavily optimized DEFLATE/zlib/gzip library
libdeflate compress better and has faster decompression than igzip.
See the silesia single core in-memory benchmark here [1] comparing zlib,libdeflate,igzip,...
https://github.com/powturbo/TurboBench/issues/4
-
Intel QuickAssist Technology Zstandard Plugin for Zstandard
- https://github.com/powturbo/TurboBench/issues/43
[1] https://github.com/powturbo/TurboBench
-
Variation on RLE to Achieve Lossless Compression for Tabular Data
Compressesing your sample file, we get 823 bytes with brotli
Download TurboBench and make your own tests:
[1] - https://github.com/powturbo/TurboBench
-
Data Compression Drives the Internet. Here’s How It Works
- igzip 1,2 is best for very fast networks > 10MB/s
brotli bring little value at decompression for users
[1] https://github.com/powturbo/TurboBench
[1] https://sites.google.com/site/powturbo/home/web-compression
[2] https://encode.su/threads/2333-TurboBench-Back-to-the-future...
-
Pigz: Parallel gzip for modern multi-processor, multi-core machines
Build or download TurboBench [1] executables for linux and windows from releases [2] ans make your own tests comparing oodle,zstd and other compressors.
[1] https://github.com/powturbo/TurboBench
[2] https://github.com/powturbo/TurboBench/releases
What are some alternatives?
pigz - A parallel implementation of gzip for modern multi-processor, multi-core machines.
QAT-ZSTD-Plugin
DirectStorage - DirectStorage for Windows is an API that allows game developers to unlock the full potential of high speed NVMe drives for loading game assets.
libdeflate - Heavily optimized library for DEFLATE/zlib/gzip compression and decompression
QATzip - Compression Library accelerated by Intel® QuickAssist Technology
parquet-format - Apache Parquet
lib842
nvcomp - Repository for nvCOMP docs and examples. nvCOMP is a library for fast lossless compression/decompression on the GPU that can be downloaded from https://developer.nvidia.com/nvcomp.
pixz - Parallel, indexed xz compressor
FPC - FPC - Fast Prefix Coder