[A83] Re: Compression update
[Prev][Next][Index][Thread]
[A83] Re: Compression update
Henk Poley wrote:
> Darren? How far have you got with your compressor?
>
> I'm currently looking a bit around in compression-land, Igor Pavlov (the
> creator of 7-zip) said to me that the deflate algorithm is possibly the
> best thing to go for. Especialy if you only need to decompress on-calc.
I agree that for decompression only, deflate would probably yeild the
best results for the average case. IIRC, deflate uses both LZ and
Huffman compression (hence the good compression ratio), but the extra
overhead incurred by implementing both algorithms almost certainly make
calc-based compression redundant, since we we be compressing at most ~24k.
Incedently, some modifications I made to the data stream format results
in my compressor beating out deflates LZ algo every so slightly in most
files (I'm counting bits, here :)
I don't know if you saw my post about ZPic83, we can be found at
http://home.iprimus.com.au/qarnos/zpic83/zpic83.zip.
It contains an LZSS decompression routine I wrote, along with a picture
compressor (PCX). I am currently re-writing a calc compressor due to the
changes I made to the algorithm which made the old one rather redundant ;)
But I can warn you that for high compression (equal to that of the
PC-based compressor), this routine will be *very* slow. I am currently
working on a couple of tricks to speed it up a bit, but if you need fast
compression the best bet will be to turn extra compression off!
Doing things like self-extracting programs may also be tricky, do to the
limited RAM. On the 83+ calcs, the compressed file could be stored (or
swapped) to the archive, but this would be of no use to 83- users.
I am really starting to get jealous of 86 owners, with their fancy
scratch RAM. Bah!
References: