UPC++ Users BOF

The Pagoda team at LBNL hosted a "Birds of a Feather" session for users of UPC++.

When: Thu Feb 16, 11am Eastern / 8am Pacific / 5pm Central European Time (90 min duration)

This 90-minute event included brief "lightning talks" by a group of UPC++ application developers who describe their experiences with UPC++:

• Paul H. Hargrove (Lawrence Berkeley National Laboratory)
  • Introduction and UPC++ overview
• Yakup Koray Budanaz (Technical University of Munich (TUM))
  • "An Adventure about Serializing: Implementing Dynamic Workload Balancing with Actors on UPC++"
• Johnny Corbino (Lawrence Berkeley National Laboratory)
  • "UPC++'s Crucial Role in Quantum Chemistry"
• Rob Egan (Joint Genome Institute)
  • "Using UPC++ to Assemble Exascale Metagenomes"
• Basilio B. Fraguela (Universidade da Coruna, A Coruna, Spain)
  • “UPC++ DepSpawn: A library for data-flow computing with hybrid parallelism”
• Sayan Ghosh (Pacific Northwest National Laboratory)
  • "Towards a more asynchronous and productive graph analytics using UPC++"
• Kirtus G. Leyba (Arizona State University)
  • "Using UPC++ and GPUs for Accelerating a Within-Host Model of Viral Infections"
• Daniel Waters (Lawrence Berkeley National Laboratory)
  • "A Case Study in Portability of UPC++ Memory Kinds"

This was followed by open Q&A in which the audience was encouraged to interact with the application developers and with the UPC++ maintainers.

UPC++ is a C++ library providing classes and functions that support Partitioned Global Address Space (PGAS) programming. The UPC++ API offers low-overhead one-sided Remote Memory Access (RMA) communication and Remote Procedure Calls (RPC), along with futures and promises. These constructs enable the programmer to express dependencies between asynchronous computations and data movement. UPC++ supports the implementation of simple, regular data structures as well as more elaborate distributed data structures where communication is fine-grained, irregular, or both. The library's support for asynchrony enables the application to aggressively overlap and schedule communication and computation to reduce wait times.

UPC++ is highly portable and runs on platforms from laptops to supercomputers and is installed at various HPC centers including NERSC, ALCF and OLCF. As a C++ library, it interoperates smoothly with existing numerical libraries and on-node programming models (e.g., OpenMP, Kokkos, CUDA, HIP).

This event was sponsored by the Exascale Computing Project, as part of the 2023 ECP Community BOF days virtual symposium.