Four pages that go one notch underneath the proof system — the CPU model a ZisK program actually runs against, how the RISC-V binary you compile becomes the ZisK ROM the machine uses, the math layer that turns an execution into a constraint system, and the hard upper bounds the current setup imposes.
Deeper
ISA & processor
The contract and the machine — the four-register c = op(a, b) shape, the source/store modes, the execution loop, and the opcode catalogue.
Read more →Deeper
Transpilation
How a standard RISC-V ELF becomes ZisK ROM at load time, instruction alignment, and the address map the result lives in.
Read more →Deeper
Arithmetization
How an execution becomes a system of polynomial constraints — AIRs, bus interactions, the three argument types, and the LogUp reduction.
Read more →Reference
Limits
The hard upper bounds the current ZisK setup imposes — input size, ROM, RAM, trace, max steps — with Ethereum-block validation as a yardstick.
Read more →Paper
ZisK Whitepaper
One notch deeper than these pages — the formal soundness arguments, polynomial-degree bounds, the exact STARK construction, and the LtHash security reductions. The full mathematical specification, opened directly from the official repo.
Read more →If you arrived here from How ZisK works, these pages fill in the what gets proved against what underneath the proof-system topology you've already seen.