Thread: BALZ v1.05 is here!

OK, a new version is out! This one features a heavily improved match finder (HC5), plus some additional parsing tricks.

http://encode.ru/balz/index.htm

Will made some break in development, after, I'm planning to do some serious changes in BALZ, just my fingers crossed around SS/Optimal Parsing. In addition, I recalled my successful LZP-driven compressor QUAD v1.04a. Just test it and see how modern LZP can perform. So I'm looking into LZP side also.

P.S.
Now you can even attach files!

Thanks Ilia!

Mirror: Download

Quick test...

BALZ [e]

A10.jpg > 844,147
AcroRd32.exe > 1,457,272
english.dic > 1,088,475
FlashMX.pdf > 3,750,407
FP.LOG > 896,154
MSO97.DLL > 1,902,127
ohs.doc > 832,897
rafale.bmp > 1,093,427
vcfiu.hlp > 733,042
world95.txt > 644,095

Total = 13,242,043 bytes


BALZ [ex]

A10.jpg > 844,110
AcroRd32.exe > 1,430,594
english.dic > 1,094,502
FlashMX.pdf > 3,737,938
FP.LOG > 856,516
MSO97.DLL > 1,875,286
ohs.doc > 824,912
rafale.bmp > 1,075,973
vcfiu.hlp > 699,565
world95.txt > 608,356

Total = 13,047,752 bytes

Will do more tests later tonight...

Originally Posted by LovePimple

Will do more tests later tonight...

Thank you!

Another quick test...

Test machine: AMD Sempron 2400+ with Windows XP SP2 operating system

Test File: ENWIK8 (100,000,000 bytes)

Timed with XTIME v1.0


BALZ [e]

Compressed Size: 30,299,042 bytes

Compression Time: 1179.954 Seconds

Decompression Time: 28.190 Seconds


BALZ [ex]

Compressed Size: 29,272,713 bytes

Compression Time: 2996.026 Seconds

Decompression Time: 25.894 Seconds

A question. What if I combine QUAD, LZPM and BALZ? The answer is a new BALZ v1.06. It can be regarded as the LZPM v2 or QUAD v2... Anyway, it's a ROLZ-based compressor featuring an Optimized Parsing in max mode and Advanced Lazy Matching with default mode. This is a crazy thing - fast compression, high compression ratio, low memory usage at compression - ~64 MB... It will be released within a week...

Of course, it easily beats LZPM, QUAD and current BALZ at SFC, being as fast as QUAD!
Furthermore, on some files its performance is truly amazing:

fp.log -> 581,503 bytes
english.dic -> 797,618 bytes
vcfiu.hlp -> 668,959 bytes
...