pkm@pierce.UUCP (P.K. Maheshwari) (02/19/86)
TO WHOM IT MAY CONCERN ---------------------- I am in need of a packing algorithm which works better than the PACK Utility of UNIX. I have also looked at COMPRESS ( developed at University of Utah ). COMPRESS works great if distinct number of input bytes is small. But if the distinct input bytes reach 256 ( binary data ), PACK works better than COMPRESS. With PACK I am getting a saving of 20-25%. If anybody has an algorithm that would do better in packing "load modules", I would like to know about it. THNX...........Padam Name: Padam Maheshwari Phone: (201) 580-4552 Organization: AT&T Bell Labs Address: 3ND15 184 Liberty Corner Rd Warren, NJ 07060-0908 NET Mail Address: ihnp4!pierce!pkm
thomas@utah-gr.UUCP (Spencer W. Thomas) (02/20/86)
Actually, compress works best if there are repeated SEQUENCES in the
input data. It doesn't really matter so much how many different codes
there are, except that one is more likely to get repeated sequences if
there are fewer codes.
--
=Spencer ({ihnp4,decvax}!utah-cs!thomas, thomas@utah-cs.ARPA)guy@sun.uucp (Guy Harris) (02/20/86)
> I am in need of a packing algorithm which works better than the > PACK Utility of UNIX. I have also looked at COMPRESS ( developed > at University of Utah ). COMPRESS works great if distinct number > of input bytes is small. But if the distinct input bytes reach > 256 ( binary data ), PACK works better than COMPRESS. With PACK > I am getting a saving of 20-25%. If anybody has an algorithm > that would do better in packing "load modules", I would like to > know about it. Are you sure you have the latest version of "compress"? I tried "pack"ing and "compress"ing "/usr/bin/suntools" (which is a BMF executable image) and "compress" did significantly better than "pack" did (it took significantly more time doing it, but that's life). Remember, Lempel-Ziv compression will discover frequently-occurring sequences of bytes, and unless your machine has all one-byte instructions you're likely to get some multi-byte sequences occurring frequently. -- Guy Harris {ihnp4, decvax, seismo, decwrl, ...}!sun!guy guy@sun.arpa (yes, really)