M03 alpha

[michael maniscalco]

An alpha release of M03 is now available (finally).

M03 is a full order context based compression algorithm for the Burrows/Wheeler Transform. This alpha release uses approximately 6N for encoding and 5N for decoding. Compression ratios and encode/decode speeds are very competitive however this release is just the most basic form of the algorithm and there is still a lot of room for improvements on speed, memory and compression.

[UPDATE 10/9/09]
Version 0.2 alpha is ready. Fixes some of the stupid errors in 0.1 alpha.

http://www.michael-maniscalco.com/do...s/M03.0.2a.zip

- Michael Maniscalco

[joerg]

sadly the program crashes on my computer
the output-file was not created
---
D:\M03>m03 e 16 db3.dmp d3-16
M03 Alpha - Context based BWT compressor
Michael A Maniscalco - 10.8.2009
email: michael@michael-maniscalco.com

Encoding bytes 0 -> 16
This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

---
Microsoft Windows (VistaI
M03.exe funktioniert nicht mehr
---
Problemereignisname: APPCRASH
Anwendungsname: M03.exe
Anwendungsversion: 0.0.0.0
Anwendungszeitstempel: 4acde538
Fehlermodulname: M03.exe
Fehlermodulversion: 0.0.0.0
Fehlermodulzeitstempel: 4acde538
Ausnahmecode: 40000015
Ausnahmeoffset: 0001640c
Betriebsystemversion: 6.0.6001.2.1.0.256.6
Gebietsschema-ID: 1031
Zusatzinformation 1: b50a
Zusatzinformation 2: f884f32182c8dc6324d444449a731ae7
Zusatzinformation 3: 2d43
Zusatzinformation 4: 188500dc0e96576c870dab71306470f8
---

[PiPPoNe92]

it works on my computer.

[michael maniscalco]

sadly the program crashes on my computer
the output-file was not created

Since it's just an alpha release, it doesn't check to see if the output file was successfully created before attempting to write to it. Perhaps there is some sort of permissions issue with creating the file on vista?

Also, a blocksize of 16 bytes isn't going to get very good compression results.

- Michael

[Matt Mahoney]

Decompression of enwik9 failed. enwik8 successfully compressed and decompressed with block size 333333334 (size 20713389). Test machine is a 2 GHz T3200, 3 GB, Vista (32 bit) Gateway laptop.

C:\tmp>timer m03 e 250000000 \res\enwik9 enwik9.m03

Timer 3.01  Copyright (c) 2002-2003 Igor Pavlov  2003-07-10
M03 Alpha - Context based BWT compressor
Michael A Maniscalco - 10.8.2009
email: michael@michael-maniscalco.com

Compressed 1000000000 -> 173944561
Time: 891.861 seconds
Done

Kernel Time  =     3.400 = 00:00:03.400 =   0%
User Time    =   863.590 = 00:14:23.590 =  96%
Process Time =   866.991 = 00:14:26.991 =  97%
Global Time  =   892.170 = 00:14:52.170 = 100%

C:\tmp>timer m03 d enwik9.m03 enwik9

Timer 3.01  Copyright (c) 2002-2003 Igor Pavlov  2003-07-10
M03 Alpha - Context based BWT compressor
Michael A Maniscalco - 10.8.2009
email: michael@michael-maniscalco.com

Decoding bytes 0 -> 250000000
This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

Kernel Time  =     0.982 = 00:00:00.982 =   0%
User Time    =    99.310 = 00:01:39.310 =  89%
Process Time =   100.293 = 00:01:40.293 =  90%
Global Time  =   110.823 = 00:01:50.823 = 100%

C:\tmp>dir
 Volume in drive C is OS
 Volume Serial Number is 66E6-426E

 Directory of C:\tmp

10/08/2009  12:57 PM    <DIR>          .
10/08/2009  12:57 PM    <DIR>          ..
10/08/2009  12:36 PM        20,713,387 enwik8.m03
10/08/2009  12:57 PM                 0 enwik9
10/08/2009  12:56 PM       173,944,569 enwik9.m03
10/08/2009  09:12 AM           182,272 M03.exe
10/08/2009  12:32 PM            95,190 M03.zip
10/08/2009  09:00 AM               314 readme.txt
               6 File(s)    194,935,732 bytes
               2 Dir(s)  28,382,105,600 bytes free

Also, actual size is 8 bytes larger than reported.

[michael maniscalco]

Version 0.2a fixes the crash on enwik9. Was actually a bad allocation causing an exception during unBWT.

Several other minor fixes as well including correcting the reported compressed size which was including the size of the internal file header in 0.1a

- Michael

[Nania Francesco]

Thanks Michael! good job!