X @Polyhedra
Polyhedraยท2025-07-25 13:00
2/- Improved shared memory across multi-threaded processes- Flexible SIMD config for better parallelism- Refined PCS interface and efficient multi-claim merging- Lower memory footprint for zkML (e.g. VGG <8GB)- Fine-grained CPU resource control and deterministic proofs- Setup, proving, and verification now cleanly separated ...