Released the first version of my open-source RTL simulator

Nov 21, 2026 Last reply: 22 hours ago 3 Replies
Jump to replies (3)

So after a lot of late nights I finally put out the first version of a new RTL simulator I've been building, and it's up on GitHub now as open source. Released it last week. Fair warning up front: I made heavy use of AI to get this thing written, which honestly is part of why it exists at all. It's called Enzan. It handles Verilog, VHDL, and mixed language designs, and there's an interactive GUI for poking at waveforms and signals instead of squinting at a console. It also hooks into Python's cocotb, which was a must-have for me since that's how I like writing testbenches. Repo is here if anyone wants to kick the tires:

formatting link
To prove it does something useful, I ran an AES testbench (tb_aes) through it. The waveforms look right and the transcript shows all test cases completed successfully, which felt pretty good to see. If you've got designs sitting around, give it a spin and see how it holds up.


I built a project the same way a while back, leaning heavily on AI to get the code written, so I know the feeling of finally pushing the release out. I have to be honest though, I ended up on the same side as the skeptics here. The question I couldn't get past with my own thing was what it actually offered over Verilator, wich most people allready have in thier flow. Without benchmarks or a side-by-side feature comparison, I found people just weren't intersted, and I can't blame them for that. Running one AES testbench with all cases passing is a nice moment, but it only shows the happy path works. What I'd want to see before trusting any simulator, including mine, is validation against the test cases other simulators already use, with outputs compared side by side. That's how you'd catch differences in event ordering and corner cases that a single demo never touches. The other concern is a fair one too. A tool people build designs around needs a commitment to years of maintenace, and an AI-generated codebase makes people wonder about that, plus the provenance question about wether any code traces back to Verilator itself. A clear statement on licensing and long-term support would go a long way.

I want to push back a bit on the tone of this thread, because the angle people keep missing is the mixed language part. My day job involves a codebase that is half Verilog and half VHDL, and the open source options I know of basically force you to pick one or jury-rig two tools together. That alone is a reason this could earn a spot, benchmarks or not. The GUI is the other thing. I have spent years squinting at console transcripts, and looking at your photo, that waveform viewer with the cursors and the hex values on the traces is exactly the part I would actually use. Being able to click through the hierarchy tree and see signal values in the Objects panel without dropping to a terminal is a real workflow change for someone like me.

Same here on the cocotb side, by the way, my testbenches are all Python, so that integration is what would make me try it at all. One question, since you have already got tb_aes running: did the mixed language support come free, or did you have to build the two language front ends separately and glue them? That detail would tell me a lot about where the hard parts of this project actually live.

For me the cocotb hook is the part that decides whether I'd even try it. My own workflow moved to Python testbenches a couple years back and I never looked back, so any new tool that doesn't plug into that is dead on arrival for me regardless of how it benchmarks. You made that a first-class feature rather than an afterthought, which is the right call. I did have a look at your screenshot, and the layout with the hierarchy tree next to the objects list feels like the viewers I grew up on, which will help people feel at home. One question though: how does the GUI handle signals from the Python side, since those aren't in the design hierarchy?

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required