Parallelism in a Region Inference Context

Explore a groundbreaking 18-minute video presentation from PLDI 2023 on combining region inference with parallel programming techniques. Delve into the implementation of a fork-join parallel construct suitable for region inference-based compilers, presented by Martin Elsman and Troels Henriksen from the University of Copenhagen. Examine the minimal higher-order language incorporating this parallel construct, including typing rules and dynamic semantics, and understand the demonstration of type soundness. Learn about the novel effect-based region-protection inference algorithm and its benefits and limitations. Discover the efficient implementation embedded in the MLKit Standard ML compiler and evaluate its effectiveness through parallel benchmarks, showcasing how this technique utilizes multi-core architectures in a higher-order functional setting. Access supplementary materials, including available and reusable artifacts, to further enhance your understanding of this innovative approach to parallelism in region inference contexts.

[PLDI'23] Parallelism in a Region Inference Context