Rendered at 09:54:06 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
pjc50 25 minutes ago [-]
This is also very well documented. Probably as part of the assessment, but it's nice to see.
It doesn't appear to have any kind of interrupts, which is quite a limitation for actual usecases, but also makes the architecture much simpler.
The use of dual phase clocking is interesting. The document describes it as having the control and data paths operating on opposite phases. I'm curious as to where you got this technique from, since it's not common (apart from the use of both edges by DDR RAM). I also suspect that it would go away if you had better tooling for managing setup and hold violations (does logisim do that for you, or did you have to manage it manually somehow?). Not all FPGA tools like nonstandard clock architectures.
sehugg 18 minutes ago [-]
The dual phase clocking is perhaps inspired by the 6502.
gsliepen 2 hours ago [-]
Very nice. I wonder if implementing a one-instruction set computer (for example something that implements Adrian Cable's subleq VM, see https://www.ioccc.org/2025/cable/) would be educational and whether it can make the design of a computer from discrete logic chips simpler or more complex. Though it would very likely not be as efficient.
gabrielsroka 9 hours ago [-]
Ben Eater's SAP came from "Digital Computer Electronics" by Albert Paul Malvino and Jerald A. Brown
I did something similar for a school capstone project. It brings back memories. Writing Verilog, working with FPGAs, that sort of thing. But this goes even further and actually gets into hardwired implementation, which is really impressive. Actually, using if else statements just creates MUXes, so you don't have fanout issues to worry about. But for something like this, you would have to handle timing calculations for rising and falling edges. It is really remarkable.
peterus 9 hours ago [-]
Awesome project, re fpga implementation one option you might want to explore are used Bitcoin miner control boards if you want the best logic units/$ ratio. I've used the EBAZ4205 (zynq 7010) control boards with a cheap/generic FT2232HL dev board and it works great. Of course it's a bit more of a pain compared to a regular dev board
wildzzz 8 hours ago [-]
If they are taking digital design classes, they'll probably be given something like a DE0-Nano that can run this just fine.
bcjdjsndon 30 minutes ago [-]
That ipc lol were all 8bit CPUs less than1 instruction per cycle?
osigurdson 7 hours ago [-]
A 2nd year project back in the day was to build a 4 bit CPU on a breadboard. We had the advantage of having an ALU IC but was still quite tough to get working!
tralarpa 5 minutes ago [-]
You were in good company. The Xerox Alto CPU used four 74181 ALUs.
chrisakoury 2 hours ago [-]
Logisim was the shit back in Uni
Very implressive tbh
You should include a screenshot of the logisim diagram
momoraul 9 hours ago [-]
the rom-to-ram bootstrap is a nice touch. after it hands over, what stops
the bootloader from writing to I-SRAM again? a mode flag in the control matrix?
hahooh 2 hours ago [-]
amazing 2nd year i was playing starcraft all day
dreamcompiler 7 hours ago [-]
"This makes the machine transparent in a way that microcode-based designs cannot be."
Every output bit m of microcode can be equivalently expressed as a logic function of n inputs where the microcode has n incoming address lines. This no less transparent than pure logic if you know the contents of the microcode. Microcode is often preferred because changing it is much easier than changing a bunch of gate logic. IMHO factoring your design into registers vs. control signals and putting the control signals into microcode makes the design more transparent than having a giant sea of gates.
HerbManic 9 hours ago [-]
With all the talk about developers being lazy and younger folk not understanding the technology they use, it is always great to see examples of core foundational work still being done. Makes the rest of that talk feel like fear mongering.
wl 6 hours ago [-]
I did an undergraduate electrical engineering degree some years ago. Building a CPU much like this one was the final project in our second digital design class. The difference here from the usual approach, as they point out, is that this is a gate-by-gate design you could assemble out of 74-series logic gates on a breadboard rather than the more common Verilog/VHDL designs that target FPGAs. Definitely a more tedious, time consuming approach, but I'm not convinced it's more conceptually difficult. By the time you're building a CPU, you've probably already built the component parts like the adders out of pure logic gates, anyway.
Not to say that this is easy, but I think you'd find that a whole lot more people are doing this kind of thing than you might think.
andrewvu0203 4 hours ago [-]
[flagged]
roshiya 2 days ago [-]
[dead]
hypfer 5 hours ago [-]
Neat!
Now do it the other way round and make the 8 bit cpus become a 2nd year EE student.
It doesn't appear to have any kind of interrupts, which is quite a limitation for actual usecases, but also makes the architecture much simpler.
The use of dual phase clocking is interesting. The document describes it as having the control and data paths operating on opposite phases. I'm curious as to where you got this technique from, since it's not common (apart from the use of both edges by DDR RAM). I also suspect that it would go away if you had better tooling for managing setup and hold violations (does logisim do that for you, or did you have to manage it manually somehow?). Not all FPGA tools like nonstandard clock architectures.
https://en.wikipedia.org/wiki/Simple-As-Possible_computer
Very implressive tbh
You should include a screenshot of the logisim diagram
Every output bit m of microcode can be equivalently expressed as a logic function of n inputs where the microcode has n incoming address lines. This no less transparent than pure logic if you know the contents of the microcode. Microcode is often preferred because changing it is much easier than changing a bunch of gate logic. IMHO factoring your design into registers vs. control signals and putting the control signals into microcode makes the design more transparent than having a giant sea of gates.
Not to say that this is easy, but I think you'd find that a whole lot more people are doing this kind of thing than you might think.
Now do it the other way round and make the 8 bit cpus become a 2nd year EE student.