BenX: Resource-Sharing Permutations for Computational Integrity
BenX, a new permutation hash, matches Poseidon2 FPGA throughput while improving hardware sharing and some zero-knowledge proving costs.
BenX is a permutation-based hash for computational-integrity proofs, designed for fast CPU execution, FPGA acceleration, and compact arithmetic circuits. Its permutation converts a Benes network into an invertible function over a prime field using a Dickson polynomial. Fully pipelined on FPGA, it matches Poseidon2 throughput and cuts latency by 1.36x on Goldilocks and 2.15x on BabyBear, while keeping shared field multipliers fully utilized. In zero-knowledge proofs it proves Goldilocks permutations 6-10x faster than Monolith and uses about 36% fewer BabyBear trace cells than Poseidon and Poseidon2, though a fast variant needs 1.3-2.6x Poseidon2 prover time.
- BenX builds an invertible prime-field permutation from a Benes network and Dickson polynomial.
- FPGA throughput matches Poseidon2, with lower Goldilocks and BabyBear latency.
- Shared NTT and hash multipliers stay fully busy, versus 6-20% for Poseidon partial rounds.
- Goldilocks permutations prove 6-10x faster than Monolith, with fewer BabyBear trace cells.
From the original article
1 min read at arxiv.orgCryptographic hash functions over integers modulo a prime play a decisive role in the efficiency and security of proof systems for computational integrity. Early designs focused on compact arithmetic circuits and efficient software execution, primarily targeting general-purpose CPUs rather than hardware accelerators.
By Luca Campa · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.