Tackling the Challenges of High-Performance Graph Analytics at Compiler Level

Tackling the Challenges of High-Performance Graph Analytics at Compiler Level

Explore COMET, a Domain-Specific Language (DSL) and compiler framework for dense and sparse tensor algebra, in this 16-minute conference talk from ACM SIGPLAN. Learn how COMET addresses challenges in scientific, data analytics, and AI workloads used at National Laboratories through progressive lowering techniques. Discover how the framework generates efficient code for heterogeneous systems by employing domain-specific and architecture-specific optimizations. Gain insights into the challenges and opportunities for optimizing high-performance data analytics at the compiler level, with a focus on graph algorithms. Understand the specific optimizations and code transformations utilized by COMET to enhance performance in graph analytics.