Advanced Codec Pack - engine list of changes


  • conexware

    Here is the list of changes for Advanced Codec Pack engine

    13-12-2016 09:56 vE4

    • zstd2 enc. progress reporting fix
    • plzma4 progress fix
    • plzma4 buffering changes

    21-12-2016 00:10 vE5

    • initial version of x64flt4

    25-12-2016 06:39 vE6

    • x64flt4 update
    • single compressed stream instead of 3
    • rc speed opt
    • compression improvement
    • rc flush after 256k bytes without any addr output
    • x64comp filter, for addr stream compression with bcj2/x64flt3

    25-01-2017 10:11 vE7

    • mpzapi filter

    potential support for external executables as .pa filters
    potential support for executables that don’t work with stdin/stdout (via winapi hooks)

    26-01-2017 13:18 vE7

    • lepton filter, same exewrap lib
    • bugfix: “Data error” with mpzapi when extracting non-solid archive

    27-01-2017 07:48 vE7

    • lepton fallback: files are now stored if lepton quits without writing anything
    • bugfix: lepton inputs that can’t be correctly restored are now reported during compression, not decompression
    • bugfix: remove mpzapi.exe crashes during extraction

    31-01-2017 07:44 vE7

    • packmp3 support

    10-02-2017 20:00 vE7

    • plzma MT bugfix

    12-02-2017 22:16 vE8

    • lepton fix to use two output streams; s0=fallback, s1=lepton
    • lepton fix to use chunksize param for fallback; lepton:c=400M uses 400M inpbuf

    13-02-2017 08:47 vE8

    • bsc support added (as “bsc3”), with :c#M,:x0,3-6,:a1/2 as params

    13-02-2017 23:39 vE8

    • bsc3: added lc param - lc0-lc2 means cf/cp/ca, lc4 means -r; (lc6 = -ca -r)

    20-02-2017 12:19 vE8

    • plzma4: 32-bit outpos bugfix
    • plzma4: loop_enc EOF check fix

    26-02-2017 19:51 vE8

    • bwt1/qlfc filters added

    27-02-2017 05:30 vE8

    • divsufsort.dll rebuilt with gcc

    28-02-2017 04:46 vE8

    • bwt1: chunksize bugfix
    • bwt1/qlfc: disable chunksize alignment to 1M

    01-03-2017 22:26 vE8

    • added qlfc2:mt=#:c=# - qlfc with integrated MT wrapper

    03-03-2017 14:58 vE9

    • updated qlfc2/MTwrap
    • added bwt2:mt=#:c=#

    08-03-2017 06:56 vE9

    • BUG: plzma4 decoder memory leak
    • BUG: workaround for divsufsort’s inverse_bw_transform doing nothing for n=1

    09-03-2017 09:31 vE9

    • BUG: 7z function CHandler::IsFolderEncrypted is buggy (outdated)
    • update bwt1 to 5N version (was 6N)
    • x64flt3: remove zero padding at the end (left from debug)

    10-03-2017 07:52 vE9

    • reflate update to ver 1l2 (bugfix)

    10-03-2017 13:41 vE9

    • BUG: ppmd_sh incorrectly parses memory size
    • BUG: ppmd_sh UpdateModel bugfix

    13-03-2017 12:08 vE9

    • added coro_init() call to deltb::Init()

    14-03-2017 17:25 vE9

    • BUG: all x64flt filters got stuck on files shorter than 8 bytes

    15-03-2017 05:57 vE9

    • 7z k_Scan_NumCodersStreams_in_Folder_MAX limit increased to 512

    17-03-2017 15:45 VE9

    • reflate speed optimization (23% faster on x64, 8% on x86)

    20-03-2017 03:30 vE9

    • BUG: lepton failed during encoding of some files; added exitcode check

    25-03-2017 04:32 vF0

    • switched default encryption to winaes

    30-03-2017 09:28 vF0

    • BUG: sometimes there’s not enough memory for winaes decrypting

    04-04-2017 19:15 vF0

    • added MTwrap-based MT zstd as zstd3 - seems incompatible with zstd2 for some reason

    06-04-2017 14:57 vF0

    • zstd3: update to 1.1.4 library

    06-04-2017 15:31 vF0

    • zstd3: fall back to zstd 1.1.0 - 1.1.4 is slower

    22-04-2017 17:42 vF1

    • plzma (plain single-threaded one)
    • bwts,bwtsh,bwtsl,bwt1h,bwt1l,cdm,cdm2

    24-04-2017 04:49 vF1

    • bwt2h,bwt2l
    • mtwrap min_chunk workaround

    06-06-2017 02:57 vF1

    • BUG: bwt2,bwt blklen=2 incorrect handling
    • mtwrap decoder buffer increased to 2*chunksize
    • mtwrap/bufring anti-MT updates
    • IC17->IC18 for x64 build

    08-06-2017 20:18 vF2

    • rep2 = rep1 + MTwrap // :c -> :d
    • added PPMD codec from original 7z (vH)

    14-06-2017 13:07 vF2

    • BUG: zstd cQuit called instead of dQuit

    15-06-2017 13:27 vF2

    • BUG: mtwrap used memcpy on overlapped memory
    • BUG: mtwrap had duplicate memcpys
    • partial buffer flush at the end of BWT2 block
    • updated version_info

    16-06-2017 09:33 vF2
    !!! all mtwrap codecs lost compatibility (rep2,cdm2,zstd3,bwt2,bwt2l,bwt2h,qlfc2) !!!

    • BUG: mtwrap handling of l=0xFFFE blocks
    • BUG: mtwrap handling of l=0x0001 blocks

    17-06-2017 12:20 vF2

    • restored vF0-compatible zstd3,bwt2,qlfc2; test scripts included
    • BUG: freezing bug is finally solved by adding dynamic buffering to MTwrap decoder
    • mtwrap decoder input buffer reduced from 2C to 1C
    • following codecs use MTwrap_v3: BWT3,BWT3H,BWT3L,QLFC3,cdm2,rep2,zstd4

    18-06-2017 23:00 vF2

    • BUG: another mtwrap freezing bug - mtwrap didn’t notice when thread with empty input quits without outputting anything
    • 32-bit variable was used for thread EOF flags, so max mt32 was supported; updated to 64.

    19-06-2017 20:09 vF2

    • BUG: freezing/data error in zstd4

    21-06-2017 16:01 vF2

    • BUG: data errors in bwt3
    • refactored mtwrap/loop_dec

    21-06-2017 22:54 VF3

    • archive cleanup
    • modded packmp3 for mp3 compression

    25-06-2017 21:11 vF3

    • ppmd_sh2 added (dX can be used instead of mem=X)
    • ppmd_sh reverted back into enc/dec template

    01-07-2017 13:02 vF3

    • alpha version of mp3det+packmp3b combo (x86 and x64 are incompatible)

    02-07-2017 21:14 vF3

    • packmp3b updated with mtwrap

    03-07-2017 06:41 vF3

    • BUG: x86 version of packmp3b crashes on decoding (problem with IC and floats; /fp:strict fixed it)
    • BUG: memory leak in packmp3b
    • BUG: crash on decoding of test4a.mp3

    04-07-2017 18:14 vF4

    • source cleanup
    • removed some experimental codecs etc

    06-07-2017 11:31 vF4

    • BUG: packmp3b formats created by 32-bit and 64-bit 7z.dll are different
    • packmp3b compression slightly retuned towards 320kbit

    16-08-2017 22:53 vF5

    • reflate2 = reflate/mtwrap; eg. reflate2:x9:c10M

    27-08-2017 16:23 vF5

    • dropped packmp3,packmp3a codecs (and corresponding .exe)
    • added lepton2 aka lepton-slow-best-ratio

    11-09-2017 07:10 vF5

    • added precomp as precomp:mt4:c64M

    12-09-2017 08:20 vF5

    • BUG: fixed precomp to not use same tempfile names in all instances
    • disabled console input in precomp on file overwrite
    • enabled ZIP/PNG/PDF/GZIP parsing in precomp
    • updated precomp handler

    13-09-2017 12:30 vF5

    • added jojpeg for jpeg compression (solid and with detection, but slow); s0=bin, s1=compressed

    14-09-2017 15:23 vF5

    • added packmp3c (2x slower than packmp3b, 1-2% better compression)

    15-09-2017 15:03 vF5

    • packmp3c bugfix (scfsi flags), a little worse compression

    19-09-2017 16:00 vF5

    • BUG: forgot coro_init for jojpeg

    19-09-2017 17:12 vF5

    • updated precomp to 0.4.6

    20-09-2017 01:37 vF5

    • jojpeg switched to gcc dlls (35% faster)

  • conexware

    @spwolf said in Advanced Codec Pack - engine list of changes:

    22-04-2017 17:42 vF1

    plzma (plain single-threaded one)
    bwts,bwtsh,bwtsl,bwt1h,bwt1l,cdm,cdm2

    24-04-2017 04:49 vF1

    bwt2h,bwt2l
    mtwrap min_chunk workaround

    New compressor for compressed data… possible 0.1%-22% gain on various already compressed data. Normal gains are more like 1%-7%, but again, this is data usually not compressible @Alpha-Testers


  • conexware

    @Alpha-Testers updated with latest list of engine changes, F3 will be in RC2


  • conexware

    updated again today.


  • conexware

    @Alpha-Testers updated with info from July 1st onwards.


  • Alpha Testers

    Nice updates! Like the jojpeg speedup.
    Waiting for PA 18.00.03 with new codecs enabled!


  • conexware

    @nikkho we are not using jojpeg yet… we will see if we end up using it, it is still too slow for something like PA.


 

6
Online

9.8k
Users

6.0k
Topics

36.7k
Posts