16

There is a wonderful project called Visual6502, and I've used it from time to time to figure things out, like exactly the difference between rti and rts. It is good to see which control lines were fired in each cycle from the Control ROM.

But Z80 is a whole other thing. It's not microcoded like the 6502 is, but instead apparently contains some finite state machines and things. I am not sure I understand why that design choice was made, especially considering the relatively more complex ISA, including variable-length instruction codes. But I really have no idea about that.

So I am interested in finding out more about how the Z80 worked. Has this chip been reverse-engineered and had its internal mechanisms explained, such that it can be understood by laypeople?

Improve this question
7
  • I am especially interested in the parts which decode instructions. My guess is that prefixes modify some kind of internal state, and all other instructions reset that state, but it would be interesting to find out more. Commented May 23, 2019 at 13:48
  • 4
    (Not enough for an answer) You may want do browse thru Ken Shirriff's Blog as he describes quite some internals for 8080/8085/Z80. (I think he's as well on RC.SE) Commented May 23, 2019 at 14:12
  • 1
    More specific link: posts tagged "Z-80" on Ken's blog Commented May 23, 2019 at 18:42
  • 2
    PLA: en.wikipedia.org/wiki/Programmable_logic_array Microcode: en.wikipedia.org/wiki/Microcode tl;dr A PLA is a piece of hardware designed to route electrical signals in a particular way to generate particular outputs for given inputs. The 6502 uses this technology to implement opcodes. Microcode is a software construct written against the hardware to implement higher-level assembler op-codes. The x86 family use this technology. They are not the same, and to call the 6502 architecture a microcoded implementation is incorrect. Commented Jun 25, 2019 at 15:50
  • 1
    An important thing to note about PLAs during the NMOS era is that many chips had only a single layer of metal, in addition to the polysilicon layer that could be used to route signals for short distances. Consequently, most of the space on a chip would be occupied not by transistors, but by wires connecting them. A PLA uses more transistors than would be needed to accomplish the same task with direct-wired logic, but minimizes the routing space between the transistors, sometimes offering a huge win. Commented Jun 19, 2020 at 20:55

2 Answers 2

Reset to default
10

Z80 has been implemented in the Visual6502 way too. However looks like the site http://www.micrologo.net/z80x is no longer operational. It was Z80 die simulation in Javascript When I searched matching stuff from mine off-line copy I found this repository:

So if you want to run the Z80 simulation, either use the online version or download the repo and open expert-z80.html file from it in your browser.

Yes a lot has been reversed especially the 4bit ALU of Z80. See Ken Shirriff's blog sites are really good and I used them while perfected my Z80 emulator:

I ended up with this ISET configuration:

Which breaks down all the instructions into their machine cycles ... Once you done that many things start to be obvious.

Improve this answer
5

The best I've seen is Goran Devic's implementation for an FPGA. He's taken the existing bits of Z-80 reverse engineering and filled in the missing pieces and connected them all together. Read through the blog and you'll find a large PNG functional diagram and a link to a bitbucket repo of all the code.

Although I haven't done a deep dive into the details, it does seem to be similar to the 6502 in that a PLA is used to decode instructions and drive various sub-circuits. It does differ at the low level in a lot of ways in how the register files and ALUs are implemented.

Improve this answer

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.