BSC STAR Group Announces New Features in Version 2024.05

May 24, 2024 — The System Tools and Advanced Runtimes (STAR) group at the Barcelona Supercomputing Center-Centro Nacional de Supercomputación (BSC-CNS) announced the release of the STAR software stack (version 2024.05), including several updates and enhancements.

Version 2024.05 of the OmpSs-2 programming model introduces new features such as directory/cache (D/C) for CUDA devices in Nanos6, as well as multiple improvements to the nOS-V task library, as well as performance global updates and bug fixes. Additionally, the enhanced LLVM/OpenMP runtime (libomp-v) now supports OpenMP free agents and detailed instrumentation via UFO. This release discontinues support for the Mercurium compiler.

Main updates of version 2024.05:

Nanos6:

Introduction of directory/cache (D/C) for CUDA devices.
New device memory allocation API for D/C managed memory.
Improvements to UFO instrumentation for improved monitoring and analysis.

nOS-V:

New batch job submission API to maximize submission efficiency.
Native mutex support (nosv_mutex_t) and obstacles (nosv_barrier_t)
Activation of Turbo mode configuration (i.e. fast-math) by default, improving floating point performance on x86-64 and aarch64 platforms.

LLVM/OpenMP (libompv):

Implemented OpenMP free agent functionality and UFO instrumentation, improving flexibility and monitoring capabilities.

LLVM/Clang:

Introduced OPENMP_RUNTIME environment variable to select preferred runtime library.
Several bug fixes and improvements to improve stability and performance.

Here is :

Add support for the OpenMP model to instrument the libompv runtime
Updates to support documentation and configuration files for better usability.
Other bug fixes and improvements

For more information and access to the repositories, please visit: https://github.com/bsc-pm/ompss-2-releases.

Recent publications:

D. Álvarez, Kevin Sala, and Vicenç Beltran, “nOS-V: Co-Executing HPC Applications Using System-Wide Task Scheduling,” 38th IEEE International Symposium on Parallel and Distributed Processing (IPDPS), San Francisco, California, United States, May 27-31, 2024
K. Sala et al, “ALPI: Enhancing Portability and Interoperability of Task-Aware Libraries”, Asynchronous Multi-Task Systems and Applications (WAMPTA), Knoxville, Tennessee, United States, February 14-16. 2024

Funding and acknowledgments:

The STAR research group is co-funded by the Zettaescale Laboratory in Barcelona, which receives funding from the Ministry of Digital Transformation and Public Services under the Resilience and Recovery Facility. The STAR research group was funded by the European Union’s Horizon 2020/EuroHPC research and innovation program under grant agreements No. 101034126 (EuPILOT), No. 955606 (DEEP-SEA) and No. ° 956416 (ExaFOAM).

Additionally, projects PCI2021-122090-2A, PCI2021-121958 and PCI2021-121961 were supported by MCIN/AEI/10.13039/501100011033 and the European Union’s NextGenerationEU/PRTR. This work was also supported by the Ramón y Cajal Program (grant number RYC2019-027592-I) and the Severo Ochoa Center of Excellence Accreditation (grant number CEX2021-001148-S), both funded by MCIN /AEI/10.13039/501100011033. . Additionally, the Departament de Recerca i Universitats de la Generalitat de Catalunya funded the BSC-UPC Programming Models Research Group (grant number 2021 SGR01007).

Source: BSC-CNS