1/
The @optimism fault proof system got a massive upgrade today:

We’re shipping MT Cannon 💥

An upgraded version of Cannon — the fault proof VM — now with 64-bit support and multithreading!

🧵...

Permissionless public infrastructure. Contributing to Optimism at OP Labs. 
🇨🇦🏔️ https://www.pauldowman.com

2/
🤔 What’s Cannon anyway?

Cannon is the fault proof virtual machine that runs the fault proof program — “op-program”. And what's op-program?

Think of it as a version of Optimism node software that runs in a reproducible way.

4/
💡 Why the 64-bit upgrade?

The original Cannon emulated a 32-bit CPU. That was a good place to start, but 32-bit CPUs are very memory-constrained.

5/
💾 64-bit = no memory limits 🚫

Modern CPUs are 64-bit and can access vastly more memory.

By upgrading Cannon to 64-bit, we completely remove the memory bottleneck.

More memory = support for larger blocks = more scalability 🚀

6/
👻 What happens when we don’t have enough memory?

If the op-program runs out of memory, the proof fails.
And if proofs fail, bad actors can get away with fraud.

7/
🧵 And why multithreading?

The op-program is written in Go. Go’s runtime uses a garbage collector — but it runs on a separate thread.

With no multithreading?
🗑️ GC doesn’t run → memory leaks → crash risk → proof fails.

8/
🧠 With MT Cannon, multithreading is enabled.
That means the garbage collector works → memory is reclaimed → system is stable even on long or heavy runs.
More efficient, more robust, more scalable 💪

9/
To sum it up:

✅ 64-bit = more memory
✅ Multithreading = stable GC
✅ MT Cannon = no more memory constraints, even for big blocks

And if you like this kind of thing: we're hiring for the proofs team at OP Labs! Get in touch if you think you might be a good fit.

3/
So to give op-program a reproducible environment we run it in Cannon, which emulates a Linux environment on a MIPS CPU. 😅