First post, by Harry Potter
Hi! I'm working on several compression techniques for several systems and am having a performance issue: the lz77 scan code is too slow. 🙁 On 8-bit computers, it is written in assembler. A 16-bit version uses some assembler, and I need to convert it all to assembler. I applied a suggestion to optimize the 8-bit version's inner loop, and it worked but only slightly. I was told about using hash-tables. I don't fully understand it. 🙁 Can somebody here explain it better for me? I have another way to optimize lz77: use an 8k array of bits, where each bit specifies whether an associated word was found or not. It helped slightly. How else can I optimize an lz77 technique?
BTW, I have a text compression for 8-bit systems called printtok. It allows you to compress strings using tokens and RLE of spaces. It helps a text adventure I'm creating's text by about 25%. Should I convert it to DOS?
Joseph Rose, a.k.a. Harry Potter
Working magic in the computer community