Scaling up HBM Efficiency of Top-K SpMV for Approximate Embedding Similarity on FPGAs

Abstract

Top-K SpMV is a key component of similaritysearch on sparse embeddings. This sparse workload does not perform well on general-purpose NUMA systems that employ traditional caching strategies. Instead, modern FPGA accelerator cards have a few tricks up their sleeve. We introduce a Top-K SpMV FPGA design that leverages reduced precision and a novel packet-wise CSR matrix compression, enabling custom data layouts and delivering bandwidth efficiency often unreachable even in architectures with higher peak bandwidth. With HBMbased boards, we are 100x faster than a multi-threaded CPU implementation and 2x faster than a GPU with 20% higher bandwidth, with 14.2x higher power-efficiency.