Browse Proceedings

Over the last few decades, the Message Passing Interface (MPI) has become the parallel-communication standard for distributed algorithms on high-performance CPUs. MPI’s minimal setup overhead and simple API calls give it a low barrier of entry, ...$$

HPC architectures are increasingly handling workloads such as AI/ML or high performance data analytics where the working data set cannot be easily partitioned, or does not fit into node local memory. This poses challenges for programming models ...

OpenSHMEM is a Partitioned Global Address Space (PGAS) style programming model for one-sided scalable communication over distributed-memory systems. The community has always focused on high levels of performance for specific communication ...

Benchmarking is an important challenge in HPC, in particular, to be able to tune the basic blocks of the software environment used by applications. The communication library and distributed run-time environment are among the most critical ones. In ...

The Partitioned Global Address Space (PGAS) programming model, OpenSHMEM, is getting more traction as a useful method for parallel programming on future-generation platforms. However, very few works have explored on the enabling of external tools ...

Endeavors to engineer the next generation of exascale platforms have resulted in a fundamental shift in system architectures. Orthogonal to what was once considered conventional wisdom, high performance systems designed today are characterized by ...

SHMEM-ML is a domain specific library for distributed array computations and machine learning model training & inference. Like other projects at the intersection of machine learning and HPC (e.g. dask, Arkouda, Legate Numpy), SHMEM-ML aims to ...

As a lightweight library-based Partitioned Global Address Space (PGAS) programming model, OpenSHMEM provides efficient one-sided and collective communications and is receiving more attention in recent years. However, task-based programming models ...

Network library APIs have historically been developed with the emphasis on data movement, placement, and communication semantics. Many communication semantics are available across a large variety of network libraries, such as send-receive, data ...

This work examines the small-message problem in OpenSHMEM applications, and proposes a new software interface that helps alleviate its detrimental performance impacts. We summarize the state-of-the-art in transport middleware that defers and ...

HPC workloads experience significant overhead due to handling network-related tasks on the CPU. Some tasks could be offloaded to a SmartNIC, thus reducing the run-time of the workload, but this typically requires explicit support in the ...

The OpenSHMEM programming model encourages application developers to partition memory into local and symmetric segments through the use of the SHMEM_SYMMETRIC_SIZE environment variable. While this can lead to improved communication efficiency, it ...