Polyhedra completes optimization of Expander framework, enhancing performance and reliability

Foresight News reported that the AI+ZK public chain Polyhedra announced it completed optimization of the Expander framework last week. Key improvements include: end-to-end integration of the GPU Prover with the Expander framework, allowing the Prover to seamlessly switch between CPU and GPU execution modes via environment variables, enabling transparent GPU acceleration without additional configuration; optimization of FPGA arithmetic primitives for Versal HBM devices, including adders and multipliers for the M31Ext3, Goldilocks, and BN254 fields, leveraging high-bandwidth memory to maximize throughput and minimize latency, laying the foundation for full hardware-accelerated proof generation on FPGA; and benchmarking of preliminary arithmetic cores to verify functional correctness and throughput scalability.