Enabling Accelerator Offload in Presto-Velox

Learn how specialized hardware acceleration enhances Presto-Velox engine performance through NeuroBlade's collaborative open-source development efforts in this 23-minute technical talk. Discover the process of identifying and addressing key bottlenecks in query execution, particularly focusing on scan and filter operations in typical workloads. Examine detailed performance measurements using the TPC-H benchmark, explore integration techniques, and understand performance profiling methodologies using tools like PBench. Follow NeuroBlade's journey in contributing these performance improvements back to the open-source community, gaining practical insights into hardware-accelerated database optimization.