Exploit Parallelism for AI Workloads with WASM and OpenMP

Explore the potential of WebAssembly (WASM) and OpenMP for optimizing AI workloads in this 25-minute conference talk by Intel experts. Delve into the challenges and benefits of implementing multithreaded functionality and vector operations in containerized environments using WASM. Learn how the wasi-threads interface, combined with OpenMP's parallel execution capabilities, can enhance performance for AI applications at the edge. Examine a practical example of a CNN Kernel to understand the complexities of integrating OpenMP's parallelization and vectorization features into WASM-based solutions.

Exploit Parallelism for AI Workloads with WASM and OpenMP - Atanas Atanasov & Andrew Brown, Intel