Commit Graph

4100 Commits

Author SHA1 Message Date
Nick Terrell
de0414b736 [libzstd] Pull CTables into sub-structure 2017-07-12 19:49:19 -07:00
Stella Lau
92bed4a7e0 [ldm] Add CHAR_OFFSET in hash function and extend header size 2017-07-12 18:47:26 -07:00
Yann Collet
d6799635a5 Merge pull request #756 from facebook/memOpt
Memory optimisations for ZSTDMT
2017-07-12 18:32:37 -07:00
Paul Cruz
7c886db0a8 changed to stderr 2017-07-12 17:28:53 -07:00
Paul Cruz
b5b18cf664 changed to malloc, added comment about adaptive compression level, and changed ternary operators 2017-07-12 17:10:58 -07:00
Paul Cruz
954d999abf fixed up freeCCtx() removed BYTE since it wasn't being used 2017-07-12 16:50:43 -07:00
Paul Cruz
3c16edd26a added copyright header, removed clean from makefile 2017-07-12 16:40:24 -07:00
Stella Lau
8de82b6eb0 [ldm] Clean up versions 2017-07-12 16:31:31 -07:00
Paul Cruz
74d3a6f5ae passes tests with adaptive compression level 2017-07-12 16:18:41 -07:00
Paul Cruz
5353d350ae working with fixed compression level and fixed dictionary size 2017-07-12 16:02:20 -07:00
Stella Lau
8ff8cdb15b [ldm] Clean up code 2017-07-12 15:12:07 -07:00
Yann Collet
33cecaaaca Merge branch 'dev' of github.com:facebook/zstd into dev 2017-07-12 14:38:36 -07:00
Yann Collet
8ef666c325 slightly increased buffer pool, to cover normal "full load" scenarios
2 buffers per active worker
+ 1 buffer for input loading
+ 1 buffer for "next input" when submitting current one
+ 1 buffer stuck in queue
2017-07-12 14:23:34 -07:00
Paul Cruz
356ddb649f working with flush job->src.size and fixed cLevel 2017-07-12 12:21:21 -07:00
Stella Lau
3a48ffd4fd Fix sumToHash to use hash space more efficiently 2017-07-12 10:53:19 -07:00
Stella Lau
e0d4162464 Minor fix for non-rolling hash 2017-07-12 09:50:24 -07:00
Stella Lau
50502519fb Switch to using rolling hash only 2017-07-12 09:47:00 -07:00
Stella Lau
583dda17a8 Update rolling hash 2017-07-11 18:13:26 -07:00
Yann Collet
052a95f77c fix : ZSTDMT_compress_advanced() correctly generates checksum
when params.fParams.checksumFlag==1.
This use case used to be impossible when only ZSTD_compress() was available
2017-07-11 17:18:26 -07:00
Paul Cruz
0a401852c4 added debug statement 2017-07-11 16:50:50 -07:00
Yann Collet
2a62f48bf4 release input buffers from inside worker thread
buffers are released sooner, which makes them available faster for next job.
=> decreases total nb of buffers necessary
2017-07-11 15:56:40 -07:00
Paul Cruz
72a183efad changed dictionary size, added debugging statements 2017-07-11 15:49:52 -07:00
Yann Collet
57236184af buffer pool : all buffers have same size
to reduce memory fragmentation.
They can be used for in or out, interchangeably.
2017-07-11 15:17:25 -07:00
Paul Cruz
7c54e09347 updated DEBUG statements 2017-07-11 15:15:41 -07:00
Paul Cruz
a3c077b8c6 added error message, updated copying dictionary into the input buffer 2017-07-11 15:00:52 -07:00
Yann Collet
34b2b95631 zstdmt : intermediate outBuffer allocated from within worker
reduces total amount of memory needed,
since jobs in queue do not have an outBuffer pre-reserved now
2017-07-11 14:59:10 -07:00
Yann Collet
16261e6951 buffer pool can be invoked from multiple threads 2017-07-11 14:14:07 -07:00
Yann Collet
fe80d45807 Merge pull request #755 from jacquesg/solaris
Solaris fix
2017-07-11 12:11:44 -07:00
Paul Cruz
34afb9b23e changed to using ZSTD_compressBegin_usingDict() and fixed strange issue with ZSTD_compressContinue() 2017-07-11 11:50:00 -07:00
Jacques Germishuys
c325c8db82 fix missing symbol 'nanosleep' for Solaris 2017-07-11 19:26:10 +02:00
Paul Cruz
7ec5928626 fixed an error where -c argument wasn't working for single files 2017-07-11 10:23:25 -07:00
Stella Lau
f6c5d07fe2 Save v3 2017-07-11 09:23:44 -07:00
Yann Collet
ef0ff7fe7f zstdmt: removed margin for improved memory usage 2017-07-11 08:54:29 -07:00
Stella Lau
6c3673f4c3 Add rolling hash 2017-07-10 22:27:43 -07:00
Paul Cruz
f918545491 made some progress on improving compression ratio, but problems exist with speed limits, and for some reason higher compression levels are really slow 2017-07-10 18:16:42 -07:00
Yann Collet
4616fad18b improved ZSTDMT_compress() memory usage
does not need the input buffer for streaming operations

also : reduced a few tests time length
2017-07-10 17:16:41 -07:00
Yann Collet
670b1fc547 optimized memory usage for ZSTDMT_compress()
Previously, each job would reserve a CCtx right before being posted.
The CCtx would be "part of the job description",
and only released when the job is completed (aka flushed).
For ZSTDMT_compress(), which creates all jobs first and only join at the end,
that meant one CCtx per job.
The nb of jobs used to be == nb of threads,
but since latest modification,
which reduces the size of jobs in order to spread the load of difficult areas,
it also increases the nb of jobs for large sources / small compression level.
This resulted in many more CCtx being created.

In this new version, CCtx are reserved within the worker thread.
It guaranteea there cannot be more CCtx reserved than workers (<= nb threads).

To do that, it required to make the CCtx Pool multi-threading-safe :
it can now be called from multiple threads in parallel.
2017-07-10 16:30:55 -07:00
Paul Cruz
01fc7c4244 changed how the detection of the last job works 2017-07-10 16:27:58 -07:00
Paul Cruz
c36552ef8a dst buffer should use ZSTD_compressBound to determine how much space it needs 2017-07-10 16:10:19 -07:00
Paul Cruz
7aa36df6df fixed memory leak that was happening when creating jobs 2017-07-10 16:03:09 -07:00
Stella Lau
ef2b728316 Clean up and refactor compress function 2017-07-10 15:48:47 -07:00
Paul Cruz
e410d63d45 made input buffer an internal part of the compression context 2017-07-10 15:37:14 -07:00
Yann Collet
3510efb02d fix : custom allocator correctly propagated to child contexts 2017-07-10 14:21:40 -07:00
Yann Collet
ee3423d709 extended fuzzer MT memory tests 2017-07-10 14:09:16 -07:00
Yann Collet
88da8f1816 fix : propagate custom allocator to ZSTDMT though ZSTD_CCtx_setParameter()
also : compile fuzzer with MT enabled
2017-07-10 14:02:33 -07:00
Yann Collet
f9524cf366 added --memtest to fuzzer 2017-07-10 13:48:41 -07:00
Stella Lau
e4155b11d7 Add warning flags to makefile and clean up code to remove warnings 2017-07-10 13:08:19 -07:00
Stella Lau
10a71d9f1c Add compression context 2017-07-10 12:38:27 -07:00
Yann Collet
e32fb0c1fe added ZSTD_sizeof_CCtx() test 2017-07-10 12:29:57 -07:00
Paul Cruz
89190ef07d renamed pool.c to poolTests.c 2017-07-10 11:32:30 -07:00