Commit Graph

52 Commits

Author SHA1 Message Date
Andy Balholm 097c1c5bc9 Use 32-bit loads in isMatch1 and isMatch5. 2020-05-15 11:17:23 -07:00
Andy Balholm 8f8b18645c Read multiple bytes in findMatchLengthWithLimit
Use 64- or 32-bit loads instead of reading a byte at a time. The
original C source did something like this, in a very C-ish way. It
needed to be simplified to translate it to Go. The exact way this works
was suggested by the assembly code in github.com/golang/snappy.
2020-05-15 10:43:19 -07:00
Erik Dubbelboer a01a7b12c9 Reuse buffers and objects using sync.Pool
This reduces the amount of garbage generated and relieves pressure on
the GC.

For a workload without reusing the Writer (using Writer.Reset) the number of allocations goes from 31 to 9.
While for a workload when you reuse the Writer the number of allocations goes from 25 to 0.
2020-05-10 10:36:19 +02:00
Andy Balholm e2c5f2109f Use len and cap instead of num_commands_ and cmd_alloc_size_. 2020-05-08 16:48:16 -07:00
Andy Balholm 4b2775ea5e Fix some ugly compound literals. 2020-05-08 15:14:47 -07:00
Andy Balholm b2497e8d72 Revert "Use sort.Sort to sort Huffman trees."
This reverts commit 6b5963335e.

It doesn't really have the performance benefit I thought it did.
2020-05-08 13:52:56 -07:00
Andy Balholm 6b5963335e Use sort.Sort to sort Huffman trees. 2020-05-07 17:51:21 -07:00
Andy Balholm 7c7a5a10ef Push output directly to dst instead of buffering. 2020-05-07 17:27:37 -07:00
Andy Balholm 625cbb6f92 Replace storage_size_ with len(storage). 2020-05-07 15:40:58 -07:00
Andy Balholm f41712f811 Remove unnecessary parameters from encodeData.
The pointers passed to out_size and output were always the same,
so there is no need to have them as parameters.

Based on
00ca10b927
and b3ee528567
2020-05-06 17:20:27 -07:00
Andy Balholm 511ca97d30 Benchmark all compression levels. 2020-05-06 16:08:01 -07:00
Andy Balholm 2c14228f02 Preserve w.commands across Reset. 2020-05-05 17:36:16 -07:00
Andy Balholm cb9be97eb7 Reuse more memory when a Writer is Reset. 2020-05-05 17:18:33 -07:00
Andy Balholm 00ca370ce2 Put full license in bench_test.go 2020-05-05 16:14:45 -07:00
Andy Balholm 3c3658f2fb A couple of tweaks 2020-05-04 16:09:27 -07:00
Andy Balholm ac459dee64 Fix excessive allocations if sortHuffmanTreeItems.
Fixes #11
2020-05-04 15:04:20 -07:00
Andy Balholm 9fc263f565 Add some benchmarks.
These are basically copied from the compress/flate package.
2020-05-04 13:35:04 -07:00
Andy Balholm b60f0d972e Copy negotiateContentEncoding.
A little copying is better than a little dependency.

Fixes #8
2019-08-21 08:13:43 -07:00
Andy Balholm ed0fd64940 Fix int overflow in test on 32-bit.
On 32-bit systems, the random array indexes were sometimes negative as a
result of converting an int64 to int.

Fixes #7.
2019-07-25 08:52:43 -07:00
Andy Balholm 71eb68cc46 Preserve compression settings when Writer is reset.
Fixes #6
2019-07-04 08:13:24 -07:00
Andy Balholm 5f990b63d2 Add README. 2019-06-21 08:47:22 -07:00
Andy Balholm 5c318f9037 Add HTTPCompressor for HTTP Content-Encoding negotiation. 2019-04-30 14:53:06 -07:00
Andy Balholm f00818cf36 Fix tests broken by API change. 2019-04-25 09:03:32 -07:00
Andy Balholm a097b55267
Merge pull request #4 from paolobarbolini/behave-like-compress
Make it behave like other compress/ packages
2019-04-25 09:01:22 -07:00
Paolo Barbolini 7ad6b2e5b1 Make brotli behave like official compress/ packages 2019-04-25 17:25:59 +02:00
Andy Balholm 580fdc1cde Removed unused error 2019-04-25 08:21:22 -07:00
Andy Balholm 1b2ddcf9b7
Merge pull request #2 from shanemhansen/shane-change-struct-scope
Lift structs to higher scope
2019-03-29 18:17:44 -04:00
Shane Hansen f6e5ca9707 gofmt 2019-03-29 15:15:07 -07:00
Shane Hansen 94370e0b46 Lift structs to higher scope
Small performance improvement (27%) by re-using search result
structs. Likely the original C version relied on stack allocation here.
2019-03-29 15:12:32 -07:00
Andy Balholm 36f6c9721d Fix warnings from go vet.
FIxes #1
2019-03-26 06:05:46 -04:00
Andy Balholm 45f5d9b05a Fix negative slice indexes. 2019-03-23 11:17:03 -07:00
Andy Balholm 08371dc992 Use const for constants. 2019-03-21 16:15:39 -07:00
Andy Balholm 52165d4974 Clean up duplicate and misplaced comments. 2019-03-21 15:34:30 -07:00
Andy Balholm 55e85f3f75 Finish un-exporting symbols. 2019-03-15 18:00:20 -07:00
Andy Balholm 4e6528a75b Remove duplicate kBlockLengthPrefixCode constants. 2019-03-15 17:27:29 -07:00
Andy Balholm 7b66c0a2f2 Un-export some more symbols 2019-03-15 17:24:40 -07:00
Andy Balholm 906c822ef2 Un-export more symbols 2019-03-15 12:05:31 -07:00
Andy Balholm 5001b3d0d7 Start un-exporting names. 2019-03-09 17:11:32 -08:00
Andy Balholm 1ce1c68433 Simplify encoder interface. 2019-03-09 16:06:37 -08:00
Andy Balholm d6c7de28d1 Simplify encoder interface to match what is used by Write. 2019-03-09 15:57:44 -08:00
Andy Balholm 99596c61a1 Consolidate composite hashes. 2019-03-09 14:01:03 -08:00
Andy Balholm ba67d8c1de Consolidate rolling hashes. 2019-03-09 13:54:31 -08:00
Andy Balholm 42dac55258 Consolidate "forgetful chain" hashes. 2019-03-09 13:34:19 -08:00
Andy Balholm c4f1bfa34f Start reducing duplication in hasher code.
The C version defined several related hasher types with preprocessor
tricks, but I split them up for the translation to Go. Now I'm
recombining them.
2019-03-09 13:01:56 -08:00
Andy Balholm 74ae18c776 Use some Go library functions. 2019-03-08 19:45:16 -08:00
Andy Balholm 7ed41e6356 Change more hasher functions to methods. 2019-03-08 16:43:15 -08:00
Andy Balholm bbbdedf380 Start converting hasher functions to methods. 2019-03-08 15:10:41 -08:00
Andy Balholm 6a14da654a Embed HasherCommon in each hasher type.
(Rather than having a pointer to the custom data in HasherCommon.)
2019-03-08 14:11:00 -08:00
Andy Balholm 7aac2143a1 Add LICENSE. 2019-03-06 17:11:20 -08:00
Andy Balholm 0da14edee1 Make BrotliDecoderState into a Reader. 2019-03-06 17:08:24 -08:00