Re[4]: [fpc-pascal] a suggestion...

2006-06-01 Thread Eduardo




I tried to benchmark a little. Archivers were limited to 512 Mb. 
Timings only in the second test. Precision is kept 1 Mb/10 seconds 
intentionally.


There are many comprehensive benchmarks, but I tested nearly the 
last versions.


OpenOffice 2.0.2 sources, 1209 Mb (there are several .gz and other 
binary files inside).
(RAR and 7z had the advantage of managing files themselves to create 
solid archives, gzip and bzip2 were used on a 1209 Mb ball made by RAR.)


Rar (3.60 beta 2):  212 Mb
gzip (1.2.4):   276 Mb
bzip2 (1.0.3):  230 Mb
7z PPMd (4.4.2):190 Mb
7z LZMA ():   too slow (~100 Kb/s, more than three hours)

After that I tried to compress FPC SVN (153 Mb), and the numbers 
were in different order (FPC ball was created in 2 minutes by Rar):


Rar 21 Mb  5 minutes
gzip29 Mb  1 minute 30 seconds
bzip2   20 Mb  3 minutes 40 seconds (is it optimized for pascal?)
7z PPMd 17 Mb  8 minutes 20 seconds
7z LZMA 16 Mb  25 minutes (well, it would be better, I believe it 
had memory problems)


I tried PPMonster archiver (open source, from the author of PPMd 
algorithm), while sources are less than 60 Kb in C, speed is often 
about 20 Kb/s. However, it promised 0.548 bits per byte of the FPC 
ball (but 1.5 for OpenOffice).


Then try one of the last compressors as PAQ8, your OO 2.0.2 
sources  will take days in compress, but i think you can get 140-150 
MB or even less and is opensource too. But it isn't used out of 
comp.compression world. Also, the WinRK3.03 is a lot faster and gets 
a bit more compression, but only works on windows. The tar.gz of 
FreeBSD (90MB decompressed) goes to 12 MB with medium compression, 
but takes 20 min in a 2GHz Athlon. Check 
http://www.maximumcompression.com/ for details and the newsgroup 
comp.compression or datacompression.info for more info.


Just change the gzip to a better zip program for make zip files, as 
7z, and you'll get an improvement from current compression, but the 
bzip2 suggestion is very good, even FreeBSD has changed the tar.gz 
system to tar.bz2 (tbz) and the difference is notable.



A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail? 


___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re[4]: [fpc-pascal] a suggestion...

2006-05-25 Thread Пётр Косаревский
 bzip2 has similar compression rates (except maybe for multimedia
 files, which isn't the case) and 7zip/LZMA usually compresses better
 than RAR.
 7zip isn't installed by default in any distro AFAIK, but at least it
 open source.
 RAR would be my last option...

I tried to benchmark a little. Archivers were limited to 512 Mb. Timings only 
in the second test. Precision is kept 1 Mb/10 seconds intentionally.

There are many comprehensive benchmarks, but I tested nearly the last versions.

OpenOffice 2.0.2 sources, 1209 Mb (there are several .gz and other binary files 
inside).
(RAR and 7z had the advantage of managing files themselves to create solid 
archives, gzip and bzip2 were used on a 1209 Mb ball made by RAR.)

Rar (3.60 beta 2):  212 Mb
gzip (1.2.4):   276 Mb
bzip2 (1.0.3):  230 Mb
7z PPMd (4.4.2):190 Mb
7z LZMA ():   too slow (~100 Kb/s, more than three hours)

After that I tried to compress FPC SVN (153 Mb), and the numbers were in 
different order (FPC ball was created in 2 minutes by Rar):

Rar 21 Mb  5 minutes
gzip29 Mb  1 minute 30 seconds
bzip2   20 Mb  3 minutes 40 seconds (is it optimized for pascal?)
7z PPMd 17 Mb  8 minutes 20 seconds
7z LZMA 16 Mb  25 minutes (well, it would be better, I believe it had memory 
problems)

I tried PPMonster archiver (open source, from the author of PPMd algorithm), 
while sources are less than 60 Kb in C, speed is often about 20 Kb/s. However, 
it promised 0.548 bits per byte of the FPC ball (but 1.5 for OpenOffice).

Naturally, there is still no good reason to change something.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal