Complex Cyber-Physical Systems (CPSs), like those used in autonomous mobility and space exploration, have an increasing need for High-Performance Computing (HPC) capabilities coupled with requirements regarding Real-Time (RT) behaviour. Parallel programming models tailored to harness the parallel capabilities of the most advanced processors efficiently are not equipped to accommodate the non-functional properties that are integral to CPSs, such as real-time behaviour (e.g., task deadlines and periods, predictability properties, and event-based execution models) and safety requirements (e.g., functional correctness, resilience, and reliability).
The HiPART project aims to provide programming models, compiler analysis and optimization techniques, and runtime systems that integrate all features needed to define and implement the critical real-time and high-performance computing requirements of complex CPSs and leverage the massive computational capabilities of modern parallel and heterogeneous architectures.
Check HiPART: https://www.linkedin.com/company/hipart
The scope of this work will bring the opportunity to develop a PhD.
Key Duties
- Parallelization, analysis and optimization of complex applications for diverse parallel and heterogeneous (GPUs, FPGAs) embedded systems
- Use, implement, and extend parallel programming models, like OpenMP, SYCL, and CUDA, to address efficiency, interoperability, adaptability, and resilience, among other things. Provide interoperability among these models and with languages from the safety-critical domain, like Rust and Ada
- Use, implement, and extend compilation frameworks, primarily based on LLVM and MLIR, targeting performance, predictability, portability, interoperability, and correctness
- Use, implement, and extend runtime systems targeting performance, heterogeneity (targeting GPUs and FPGAs), interoperability, and fault tolerance
- Write technical reports and papers
- Attend technical meetings and conferences
Data de tancament: Dilluns, 16 Desembre, 2024
Més posts de Recerca
Cap comentari:
Publica un comentari a l'entrada