433d4d30f5
T1: (failed) optimization attempt to use signed magnitude representation ( #931 )
...
Instead of using two-complement integer representation of data samples during
code-block decoding, use a signed magnitude representation to avoid comparison
branches in passes.
However the (best) timings show that it is actually slightly slower, or at
least no better. 51649 ms with this attempt vs 51536 before on MAPA_005.jp2
reencoded with default options.
Two variants with sign in msb (the default) or in lsb lead to similar
performances
2017-05-22 15:10:42 +02:00
563bd8499e
Reformat whole codebase with astyle.options ( #128 )
2017-05-09 20:46:20 +02:00
9a07ccb3d0
Add overflow checks for opj_aligned_malloc ( #841 )
...
See
https://pdfium.googlesource.com/pdfium/+/b20ab6c7acb3be1393461eb650ca8fa4660c937e/third_party/libopenjpeg20/0020-opj_aligned_malloc.patch
2016-09-15 01:57:53 +02:00
f88c9974e2
Flags in T1 shall be unsigned ( #840 )
...
This will remove some conversion warnings
2016-09-14 23:46:46 +02:00
0954bc11e3
Fix some warnings ( #838 )
...
Fix warnings introduced by uclouvain/openjpeg#786
2016-09-14 00:12:43 +02:00
48c16b2c19
Merge branch 'master' of https://github.com/uclouvain/openjpeg into tier1_optimizations_multithreading_2
...
Conflicts:
src/lib/openjp2/t1.c
2016-09-08 10:30:09 +02:00
9f24b078c7
Change 'restrict' define to 'OPJ_RESTRICT' ( #816 )
...
Visual Studio 2015 does not pass regression tests with `__restrict` so kept disabled for MSVC.
Need to check proper usage of OPJ_RESTRICT (if correct then there’s
probably a bug in vc14)
Closes #661
2016-09-06 00:49:53 +02:00
7d3c7a345f
Be robust to failed allocations of job structures
2016-05-26 23:51:32 +02:00
5fbb8b2645
Use thread-pool for T1 decoding
2016-05-25 21:02:07 +02:00
7092f7ea11
Fix MSVC210 build issue (use of C99 declaration after statement) introduced in ba1edf6cd4
2016-05-23 16:00:28 +02:00
107eb31531
Improve perf of opj_t1_dec_sigpass_mqc_vsc() and opj_t1_dec_refpass_mqc_vsc() with loop unrolling
2016-05-23 13:45:15 +02:00
8371491a99
Better inlining of opj_t1_updateflagscolflags() w.r.t. flags_stride
2016-05-23 11:53:54 +02:00
956c31d5a6
opj_t1_dec_clnpass(): remove useless test in the runlen decoding path (of the non VSC case)
2016-05-23 11:53:54 +02:00
93f7f90711
opj_t1_decode_cblks(): tiny perf increase when loop unrolling
2016-05-23 11:53:53 +02:00
1da397e94a
Tier 1 decoding: add a colflags array
...
Addition flag array such that colflags[1+0] is for state of col=0,row=0..3,
colflags[1+1] for col=1, row=0..3, colflags[1+flags_stride] for col=0,row=4..7, ...
This array avoids too much cache trashing when processing by 4 vertical samples
as done in the various decoding steps.
2016-05-23 11:53:53 +02:00
ba1edf6cd4
Reduce number of occurrences of orient function argument
...
This is essentially used to shift inside the lut_ctxno_zc, which we
can precompute at the beginning of opj_t1_decode_cblk() /
opj_t1_encode_cblk()
2016-05-22 00:50:34 +02:00
23a01dfdef
Specialize decoding passes for 64x64 code blocks
2016-05-21 16:27:35 +02:00
d8fef96f23
Improve code generation in opj_t1_dec_clnpass()
...
Add a opj_t1_dec_clnpass_step_only_if_flag_not_sig_visit() method that
does the job of opj_t1_dec_clnpass_step_only() assuming the conditions
are met. And use it in opj_t1_dec_clnpass(). The compiler generates
more efficient code.
2016-05-21 15:52:02 +02:00
c539808d09
opj_t1_updateflags(): tiny optimization
...
We can avoid using a loop-up table with some shift arithmetics.
2016-05-21 15:46:56 +02:00
e1e018a8dc
Fix UBSan gcc warning for first arg to memset non null
2016-05-02 12:13:24 +02:00
72deb588cb
Merge pull request #706 from mayeut/issue135
...
Fix issue 135
The fix is legal regarding the standard but I did not manage to find out if it covers a bug in opj_t2_read_packet_data or if the file is corrupted
2016-04-30 13:51:01 +02:00
319fc971fe
cppcheck fix for openjp2 ( #740 )
2016-04-29 23:49:17 +02:00
20789fed4e
Fix issue 135
...
dwt_interleave_h.gsr105.jp2 now has the same output as kakadu
issue399 is corrupted. Only the corrupted part changes.
Update known failures for x86 MD5
NR-DEC-kodak_2layers_lrcp.j2c-31-decode-md5
NR-DEC-kodak_2layers_lrcp.j2c-32-decode-md5
NR-DEC-issue135.j2k-68-decode-md5
2016-04-27 23:15:47 +02:00
e6881e7527
Fix negative shift left reported by UBSan ( #758 )
...
Follow-up of #757
This shall have no performance impact on 2’s complement machine where
the compiler replaces the multiplication by power of two (constant) by
a left shift.
Verified at least on MacOS Xcode 7.3, same assembly generated after fix.
2016-04-27 21:56:09 +02:00
6f2ac3ee36
Fix negative shift left reported by UBSan ( #757 )
...
This shall have no performance impact on 2’s complement machine where
the compiler replaces the multiplication by power of two (constant) by
a left shift.
Verified at least on MacOS Xcode 7.3, same assembly generated after fix.
2016-04-26 22:06:06 +02:00
ae4799ad07
Add some missing static
...
Still needs to check j2k.c & jp2.c
Update uclouvain/openjpeg#243
2015-07-18 02:39:32 +02:00
c9e231508b
various small T1 optimizations
2015-07-03 15:23:15 +02:00
e966fef3eb
[trunk] fixed a possible overflow in opj_t1_encode_cblks call to opj_int_fix_mul (fixes issue 141)
2014-12-20 13:03:11 +00:00
8d320bbdf8
[trunk] check number of components when getting mct norm (fixes issue 436)
2014-12-09 12:09:08 +00:00
c3629e37a2
[trunk] significantly reduces memory for single tile RGB encoding (fixes
...
issue 375)
2014-11-03 14:51:41 +00:00
6868ee373e
added memory allocation checks (fixes issue 355)
2014-09-19 10:26:35 +00:00
50a205d20f
[trunk] improve memory management (fixes issue 359)
2014-09-16 15:48:04 +00:00
d19a4ab676
[trunk] updated copyright and added copyright notice required by ISO, in each file; updated AUTHORS, NEWS
2014-04-03 15:30:57 +00:00
73c9e85642
[trunk] Another round of fixes for sign conversion warnings.
...
Update issue 256
2014-03-10 16:17:01 +00:00
ee3b02932a
[trunk] Another round of fixes for sign conversion warnings.
...
Update issue 256
2014-03-07 11:43:50 +00:00
7afad19c77
[trunk] Remove a simple warning about int -> float conversion
2014-03-03 16:02:36 +00:00
3c7a7a1675
[trunk] Remove deprecated functions.
...
Update issue 268
2014-03-03 08:30:36 +00:00
fcf9fa3651
[trunk] add opj_ prefix to some internal define and use the new opj type instead of int and float in some files
2012-11-16 08:29:43 +00:00
78a8d64a92
[trunk] use OPJ_INT32 instead of int in t2.c
2012-11-15 16:46:08 +00:00
ec0fe09138
[trunk] move opj_bool to OPJ_BOOL to follow the other type
2012-11-15 13:13:36 +00:00
7307bb6e9a
[trunk] remove warnings raised by flags -Wall -Wextra -pedantic and vs9 analyzer
2012-11-13 15:16:35 +00:00
cdea5c8eef
[trunk] remove some warnings from -Wall -Wextra -pedantic
2012-10-26 10:04:55 +00:00
bcc386e3bb
[trunk]remove some warnings raised by Wall
2012-10-26 09:39:52 +00:00
b63d618e48
[trunk] remove last warnings from t1
2012-10-25 17:31:18 +00:00
8864c2d59b
[trunk] rename t1_getwmsedec_v2 to opj_t1_getwmsedec and remove t1_getwmsedec
2012-10-25 17:27:22 +00:00
c75073fbf4
[trunk] cancel previous commit about use of t1_decode_cblks from v2. We should use t1_decode_cblks from 1.5 because it is faster (rev 557)
2012-10-25 17:23:54 +00:00
cf4529d058
[trunk] use the opj_ prefix for last t1 functions
2012-10-25 15:50:58 +00:00
9b6900b058
[trunk] [trunk] remove old opj_tcp and rename opj_tcp_v2 to opj_tcp
2012-10-25 15:15:02 +00:00
06615a556d
[trunk] rename opj_tcd_truct_v2 to opj_tcd_struct
2012-10-25 14:49:10 +00:00
db08494776
[trunk] rename flag_t to opj_flag_t
2012-10-24 14:06:10 +00:00
Even Rouault
Matthieu Darbois
Julien Malik
Antonin Descampe
julienmalik
Aaron Boxer
Mathieu Malaterre
Mickael Savinaud