aeff128331
change seqDef.offset into seqDef.offBase
...
to better reflect the value stored in this field.
2021-12-23 17:56:08 -08:00
e145b58cfd
changed seqDef.matchLength into seqDef.mlBase
...
since this is effectively what is stored in this field (== matchLength - MINMATCH).
This makes it clearer what needs to be done when reading from / writing to this field.
2021-12-23 13:39:46 -08:00
b77fcac61f
change ZSTD_storeSeq() interface to accept matchLength
...
instead of mlBase.
This removes the need to do `- MINMATCH` at every call site.
The new interface contract is checked with an `assert()`.
2021-12-23 12:03:33 -08:00
eebf24bd2a
Merge pull request #2947 from MehdiChinoune/patch-1
...
Fix zstd-static output name with MINGW/Clang
2021-12-22 11:13:40 -08:00
14a0eaf73b
Fix zstd-static output name with MINGW/Clang
2021-12-22 10:16:41 +01:00
41153071a0
updated manual
2021-12-21 08:52:50 -08:00
abc694f53e
update man pages
2021-12-20 14:24:06 -08:00
1219b7f3e3
updated changelog for v1.5.1
2021-12-20 12:28:27 -08:00
24960134d9
Merge pull request #2940 from facebook/revert-2885-limit-level-32bit-systems
...
Revert "Limit `ZSTD_maxCLevel` to 21 for 32-bit binaries."
2021-12-20 12:27:28 -08:00
a9e43b37d0
Revert "Limit `ZSTD_maxCLevel` to 21 for 32-bit binaries."
2021-12-20 11:43:14 -08:00
85bd1e663b
Merge pull request #2934 from facebook/lazy_rebalance
...
rebalance lazy compression levels
2021-12-20 11:37:44 -08:00
2624652a32
Merge branch 'dev' into lazy_rebalance
2021-12-20 08:28:21 -08:00
80a28f2078
update regression results
2021-12-20 07:54:57 -08:00
9fc7f38e12
added target update_regressionResults
...
to automate the creation of updated tests/regression/results.csv
2021-12-16 21:46:14 -08:00
01adddc3e0
update regression results
2021-12-16 20:43:23 -08:00
7c7b9244d6
update regression results
2021-12-16 16:07:54 -08:00
47466d2192
Merge branch 'dev' into lazy_rebalance
2021-12-16 14:46:21 -08:00
9a32492730
updated regression results.csv
2021-12-16 14:39:30 -08:00
925649b2d0
fixed version number for fast modes
2021-12-16 11:52:43 -08:00
f829c32258
forgot the chainlog is effectively a "fake" value with rowHash
...
the only value which makes sense is `hashlog-1`
as it mimics the real memory usage.
2021-12-16 11:37:40 -08:00
5c96146db3
Merge pull request #2931 from facebook/fix_rowlog
...
fixed incorrect rowlog initialization
2021-12-16 09:48:12 -08:00
5c63096269
updated benchmark for v1.5.1
...
answers #2764
2021-12-16 04:46:21 -08:00
8a724f91cb
minor changelog update
2021-12-16 01:07:15 -08:00
7a347769dc
minor changelog update
2021-12-15 21:45:30 -08:00
db1b408a2f
rebalance lazy compression levels
2021-12-15 21:33:31 -08:00
c8d6067615
fixed incorrect rowlog initialization
...
the variable has only very limited usage,
being only used once at the beginning of the block for prefetching only,
hence the error had no impact on compression ratio.
2021-12-15 14:37:05 -08:00
c5f1e826ca
Merge pull request #2925 from embg/dict_training_sample_limit_size
...
Allow user to specify memory limit for dictionary training
2021-12-15 15:58:17 -05:00
622fee6140
minor changelog update, for clarity
2021-12-15 11:56:11 -08:00
eaf786242d
Merge pull request #2929 from facebook/sse_row_lazy
...
simplify SSE implementation of row_lazy match finder
2021-12-15 11:47:15 -08:00
d5933a4009
Merge pull request #2930 from nolange/reduce_tables_to_8bit
...
Reduce tables to 8bit
2021-12-15 12:43:25 -05:00
2fbb1d10c1
Reduce bit tables to 8bit
...
This saves some 1.7Kb in rodata section (x86_64, zstd tool),
while assembler code stays the same except
the type of a few load/extend instructions.
Should not have negative performance implications.
2021-12-14 23:47:57 +01:00
99923dfc1a
Add typedefs for 8bit (un)signed
...
To make code more expressive, add U8 and S8 typedefs
2021-12-14 23:47:57 +01:00
64205b7832
Fix performance degradation with -m32 ( #2926 )
2021-12-14 15:53:50 -05:00
71c0c07c19
Allow user to specify memory limit for dictionary training
2021-12-14 14:29:01 -05:00
5e2fede604
Merge pull request #2921 from felixhandte/neg-lvl-stagger-step
...
Stagger Stepping in Negative Levels
2021-12-14 14:13:57 -05:00
05430b25a8
roll SSE implementation of row_lazy match finder
...
mostly for maintenance convenience.
Performance wise, there is very little change,
slightly faster for slog 3 & 4,
neutral or very slightly negative for slot 5 & 6.
2021-12-14 10:44:23 -08:00
450fca9704
Update Regression Tests w/ New Sizes
2021-12-13 17:29:32 -05:00
82a49c88f9
Increment Step by 1 not 2
...
I couldn't find a good way to spread `ip0` and `ip1` apart when we accelerate
due to incompressible inputs. (The methods I tried slowed things down quite a
bit.)
Since we aren't splaying ip0 and ip1 apart (which would be like `0_1_2_3_`, as
opposed to the `01__23__` we were actually doing), it's a big ambitious to
increment `step` by 2. Instead, let's increment it by 1, which has the benefit
sliiightly improving compression. Speed remains pretty much unchanged.
2021-12-13 16:59:33 -05:00
3e2a70b6fb
Merge pull request #2905 from 15596858998/dev_1205
...
add test case
2021-12-13 13:45:23 -08:00
6ca5f42402
Rewrite `step` to Track Increment Between Pairs of Positions
...
The position updates are rewritten from `ip[N] = ip[N-1] + step` to be
`ip[N] = ip[N-2] + step`. This lets us only deal with the asymmetric spacing
of gaps at setup and then we only have to keep a single `step` variable.
This seems to work quite well on GCC and Clang!
2021-12-13 14:48:26 -05:00
b8434cb754
Allow Templating `ZSTD_fast` Matchfinders on Acceleration (Lvl < -1)
2021-12-13 14:46:57 -05:00
65404fe14a
Merge pull request #2923 from IAL32/patch-1
...
typo: Small spelling mistake in example
2021-12-13 13:15:21 -05:00
c69d13eb99
Update playTests.sh
2021-12-13 08:58:42 +08:00
e0f9dc0dde
typo: Small spelling mistake in example
...
Just a couple of characters:
`main` -> `may`
2021-12-11 12:02:23 +01:00
252ef866fb
Merge pull request #2922 from facebook/x32
...
x32 compatibility
2021-12-11 00:12:10 -08:00
e1ab2200ff
fixed x32 compatibility
2021-12-10 21:02:17 -08:00
c94cda283c
added x32 compatibility test
2021-12-10 20:56:20 -08:00
ace6a7e746
Decompose `step` into Two Variables
...
This avoids an additional addition, at the cost of an additional variable.
2021-12-10 16:44:23 -05:00
22501cd283
Stagger Application of `stepSize` in ZSTD_fast
...
This replicates the behavior of @terrelln's `ZSTD_fast` implementation. That
is, it always looks at adjacent pairs of positions, and only applies the
acceleration every other position. This produces a more fine-grained
acceleration.
2021-12-10 16:44:23 -05:00
4cc5e2818a
complete changelog with #2885
2021-12-09 09:53:45 -08:00
Yann Collet
Yann Collet
مهدي شينون (Mehdi Chinoune)
Yann Collet
Elliot Gorokhovsky
Felix Handte
Norbert Lange
binhdvo
Elliot Gorokhovsky
Nick Terrell
zx123123
Adrian Castro