Don't use this alpha version for important data. Call "Test" command after each compressing operation.
What's new:
- BZip2 compressing / decompressing now can work in multi-threading mode
- Multi-threading mode now is default for multi-processor systems
- New switch: -scs{WIN|DOS|UTF-8} specifies charset for list files.
Default charset for list files is UTF-8 now.
- New switch: -mmc={N} allows to increase compression ratio for
LZMA and Deflate in some cases
- New localization: Albanian
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thanks very much for implementing -scs{WIN|DOS|UTF-8} for the list file.
Comes there also an alpha or beta command line version available?
Or is the development only in the full GUI version?
Thank again,
Willem
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Actually, there is an option to switch between fast with high memory usage and slow with medium memory usage:
"mf={MF_ID}
Sets Match Finder for LZMA. Default method is bt4. Algorithms from hc* group doesn't provide good compression ratio, but they often works pretty fast in combination with fast mode (a=0). Memory requirements depend from dictionary size (parameter "d" in table below).
MF_ID Memory Description
bt2 d*9.5 + 4 MB Binary Tree with 2 bytes hashing.
bt3 d*11.5 + 4 MB Binary Tree with 3 bytes hashing.
bt4 d*11.5 + 4 MB Binary Tree with 4 bytes hashing.
hc4 d*7.5 + 4 MB Hash Chain with 4 bytes hashing."
Note that bt2 is slower than the 4.32 version with bt4(b).
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Please test also BZip2 in new version:
1) 7z, normal, method = BZip2, solid, multi-threading = on
2) 7z, normal, method = BZip2, solid, multi-threading = off
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
No, Werner. mc stands for Match-finder Cycles. There are 1000000000 posible values so you don't want to test all of them. According to the help file a big number should be better but this is not always true.
According to my tests, the speed loss with a big mc is very small for mf=bt* but for mf=hc4 is enormous.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I guess that "-m" is the switch to set compression method parameters: to it you append the name and value of the parameter so for parameter "mc={N}" (Match-finder Cycles) you get
"-mmc={N}"
but itr is not a separate switch, just switch -m setting the mc parameter
Ddot
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- The shell extension problem (crashes in the Start Menu, with multiples files,
etc) affect all versions of 7-zip. It's really strange:
Now I have no 32-bit XP installed. Please describe simplest actions to crash.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Alpha probably is not tuned for win98se
selected 2 files, same context menu crash
- 7-zip is associated with 7z archives,
- Integrate 7-Zip in shell context menu is enabled
- Cascaded contextmenu is enabled
(if that's default enough)
Add to archive command within 7zFM.exe gives message
error starting program
The 7ZG.EXE file is linked to missing export
KERNEL32.DLL:GlobalMemoryStatusEx.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
7-Zip 4.34 alpha 1 for 32-bit Windows:
http://www.7-zip.org/alpha/7z434a1.exe
7-Zip 4.34 alpha 1 for 64-bit Windows x64:
http://www.7-zip.org/alpha/7z434a1-x64.exe
Don't use this alpha version for important data. Call "Test" command after each compressing operation.
What's new:
- BZip2 compressing / decompressing now can work in multi-threading mode
- Multi-threading mode now is default for multi-processor systems
- New switch: -scs{WIN|DOS|UTF-8} specifies charset for list files.
Default charset for list files is UTF-8 now.
- New switch: -mmc={N} allows to increase compression ratio for
LZMA and Deflate in some cases
- New localization: Albanian
Thanks very much for implementing -scs{WIN|DOS|UTF-8} for the list file.
Comes there also an alpha or beta command line version available?
Or is the development only in the full GUI version?
Thank again,
Willem
Beta will be in 3-5 days, if alpha is OK.
Still eating ram....:-| 405 mb's
With 4.32..............>369mb's
Settings: ultra-dic 32 mb's.
Dou you fix this? Or still to 4.32?
To the people who make flames!:
I don't want upgrade the ram, i think 512 (actually) it's enough.
The compressor (.33 and .34 do the same (okay, more fast)) compression but using more memory. :-|
> I don't want upgrade the ram, i think 512 (actually) it's enough.
Then use smaller dictionary, or older version. Simple. The tradeoff was memory for speed.
Unless Igor can provide the option to swith between old and updated LZMA...
Actually, there is an option to switch between fast with high memory usage and slow with medium memory usage:
"mf={MF_ID}
Sets Match Finder for LZMA. Default method is bt4. Algorithms from hc* group doesn't provide good compression ratio, but they often works pretty fast in combination with fast mode (a=0). Memory requirements depend from dictionary size (parameter "d" in table below).
MF_ID Memory Description
bt2 d*9.5 + 4 MB Binary Tree with 2 bytes hashing.
bt3 d*11.5 + 4 MB Binary Tree with 3 bytes hashing.
bt4 d*11.5 + 4 MB Binary Tree with 4 bytes hashing.
hc4 d*7.5 + 4 MB Hash Chain with 4 bytes hashing."
Note that bt2 is slower than the 4.32 version with bt4(b).
I tried the 4.34a1 vs 4.32final.
The speed it's amazing, and compress a bit more :O
I want to try to test with Open office.
Soon the results.
Some results:
I used open office 2.0.1, english version without java.
LZMA, 32mb's, search word 32, ultra, solid.
Machine:
p4 ht 2.6 512 ram
======================
4.32
6'40''365 mb's ram
58mb's
======================
4.34
4'32'' 380mb's ram
58 mb's
Not too bad :D really fast.
Please test also BZip2 in new version:
1) 7z, normal, method = BZip2, solid, multi-threading = on
2) 7z, normal, method = BZip2, solid, multi-threading = off
>New switch: -mmc={N} allows to increase
>compression ratio for LZMA and Deflate
>in some cases
Igor, mmc stands for multi media compression?. What are the possible values for N?.
Regards,
Werner
No, Werner. mc stands for Match-finder Cycles. There are 1000000000 posible values so you don't want to test all of them. According to the help file a big number should be better but this is not always true.
According to my tests, the speed loss with a big mc is very small for mf=bt* but for mf=hc4 is enormous.
The help file only talks about the mc switch, not the mmc one...
"The help file only talks about the mc switch, not the mmc one..."
Are they different?
"mc switch" = Match-finder Cycles?
"mmc switch" = ???
I guess that "-m" is the switch to set compression method parameters: to it you append the name and value of the parameter so for parameter "mc={N}" (Match-finder Cycles) you get
"-mmc={N}"
but itr is not a separate switch, just switch -m setting the mc parameter
Ddot
Tested. Some very small improvenets.
http://uclc.info
It's possible to use -mmc={N} for .zip format too.
Thanks for the info.
Will try it out.
Due to all the new options for the ZIP format it clearly outperforms KZIP on some teststets.
1) There are updated versions of the Macedonian, Italian, Czech, Slovenian and Spanish translation. See the patches tracker.
2) The shell extension problem (crashes in the Start Menu, with multiples files, etc) affect all versions of 7-zip. It's really strange:
ModVer: 3.13.0.0 Offset: 0000741a
ModVer: 4.27.0.0 Offset: 0000364d
ModVer: 4.28.0.0 Offset: 00003644
ModVer: 4.29.0.0 Offset: 00003644
ModVer: 4.30.0.0 Offset: 00001828
ModVer: 4.31.0.0 Offset: 00001828
ModVer: 4.32.0.0 Offset: 00001828
ModVer: 4.33.0.0 Offset: 0000180f
ModVer: 4.34.0.0 Offset: 0000180f
- The shell extension problem (crashes in the Start Menu, with multiples files,
etc) affect all versions of 7-zip. It's really strange:
Now I have no 32-bit XP installed. Please describe simplest actions to crash.
Thank you for all the hard work Igor. I will test it out.
Igor:
Tested with bzip of 4.34a1 (ooo 2.0.1)
7z, normal, method = BZip2, solid, multi-threading = off : 3' 03''
Size: 76 mb's
7z, normal, method = BZip2, solid, multi-threading = on: 2' 40''
Another thing, if i select HT in lzma this increases the ram, but with bzip2 no.
>Now I have no 32-bit XP installed. Please describe simplest actions to crash.
Select two or more files in Windows Explorer -> right click -> Open = Crash!!!
Start Menu -> Programs -> right click -> Open or Explore = Crash!!!
More information about this bug:
http://sourceforge.net/forum/forum.php?thread_id=1443054&forum_id=45798
Another bug (reported several times):
http://sourceforge.net/forum/forum.php?thread_id=1441242&forum_id=45797
- Select two or more files in Windows Explorer -> right click -> Open = Crash!!!
Is it for default settings of 7-Zip?
Alpha probably is not tuned for win98se
selected 2 files, same context menu crash
- 7-zip is associated with 7z archives,
- Integrate 7-Zip in shell context menu is enabled
- Cascaded contextmenu is enabled
(if that's default enough)
Add to archive command within 7zFM.exe gives message
error starting program
The 7ZG.EXE file is linked to missing export
KERNEL32.DLL:GlobalMemoryStatusEx.
Please try fixed version:
http://www.7-zip.org/alpha/7z434a1m.zip