This week’s Expander progress focused on GPU & FPGA Acceleration for Expander. Let's dive into the updates.

1/ FPGA Linear GKR Implementation: Completed the full hardware implementation of the dual-phase prepare operator in the LinearGKR algorithm, including atomic indirect operations on FPGA. This marks a major step toward a fully pipelined LinearGKR acceleration flow. RTL simulation completes.

2/ Resolve GPU's inefficiency via FPGA: Resolved the efficiency bottleneck in the random accumulation writes of the GPU prepare phase, increasing the effective bandwidth from 1 TB/s to 6 TB/s by using FPGA.

3/ BN254 Pipeline Optimization: Currently addressing the issue of excessive pipeline depth in BN254 arithmetic logic on FPGA, targeting better timing closure and resource balancing. Stay tuned for more updates.