Comparing an LZ4 Decompressor on Four Legacy CPUs

https://archive.ph/x3wgL if you want to read without finding and clicking the "no, your interest in stalking me is not legitimate" box that is nested in extra clicks for each of the partners (seriously wordpress, if you are happy to be so significantly non-compliant just be honest and don't comply at all).

Bumbershoot is a great blog if you are a 1980s home computer/console enjoyer, but were never quite sure how it all worked. He is trying to figure this out himself, so he goes step-by-step through getting his games working on a particular system.

So there are a lot of ASM 'deep dives' like this entry (which probably won't get too much traction.) My one complaint is he has a lot of content, but the blog is hard to navigate.

This was a tough but fun read. It's my understanding that the three restrictions ensure, by way of overshooting, that there's no need to do any expensive if-tests. Everything can just go full throttle.

Algorithms in the LZ family were always popular with self-decompressing executables, because of their simplicity and high efficiency; like a natural extension of RLE, they can achieve very high compression ratios while needing only a few dozen bytes for a decompressor.

This has great detailed comparisons between implementations on different 8-bit CPUs, a good study of how their differing approaches impact real code (e.g. 6502 vs Z-80) or enhanced versions of others (8080 vs. Z-80 vs. 8086).