Sep 2026· IEEE International Conference on Cluster Computing· pp. 338-349· 0 citations· 37 references
Abstract
Data processing units (DPUs)-network interface controllers (NICs) with onboard general-purpose processors-have been explored as communication accelerators, but their utility as asymmetric coprocessors for memory-bandwidth-bound scientific HPC workloads remains an open question. We investigate this question in the context of the MiniEM proxy application. Specifically, we accelerate an algebraic multigrid (AMG)-based Maxwell solver, representing a class of solvers whose work decomposes into concurrent, imbalanced subtasks. Our implementation offloads one subtask to NVIDIA BlueField-3 (BF3) DPUs using a heterogeneous SPMD MPI model, requiring no DPU-specific offload framework. We then characterize when this strategy is profitable as a function of the CPU/DPU rank ratio. On a Xeon Broadwell + BF3 cluster and an EPYC Zen2 + BF3 cluster, we measure end-to-end speedups of up to 1.3× and 1.7×, respectively. We also show that as few as 4 of a BF3's 16 cores suffice to outperform the best CPU-only configuration.
Unified Shared Memory (USM) has emerged as a promising programming model for simplifying data management in heterogeneous CPU-GPU HPC systems with the AMD MI300A APU. However, its performance implications in real-world HPC applications remain insufficiently understood. In this work, we go beyond microbenchmarking by ev...
Yi Ju, Diego Jiménez, Mou Lin et al.· eScience· 0 citations
Solving large sparse linear systems by Gaussian elimination is a challenging computational task due to the introduction of non-zero elements, or fill-in. Practitioners use heuristics such as the nested dissection reordering algorithm to try to reduce fill-in. In practice, reordering codes do not achieve the same parall...
Ji-Ya Su, François-Henry Rouet, J. Beisheim et al.· 0 citations
This work presents parallel algorithms for computing dependent partitions, and shows that these algorithms perform well on accelerators like GPUs and enable dynamic and on-line partitioning during steady-state execution.
Rohan Chanani, Alex Aiken, Michael Bauer et al.· 0 citations
Simulation models calibrated on one hardware generation are reused without validation. A SimGrid/SMPI model of Fletcher, an anisotropic reverse time migration (RTM) application, calibrated on one GPU per node at 1 Gbit/s, predicted higher bandwidths raise the computing-time fraction, untested with multi-GPU nodes. We r...
Arthur Andrade da Silva, Vinícius Daniel Spadotto, C. Künas et al.· Anais do XXVII Simpósio em S...· 0 citations
Graph partitioning is a key enabler in computational science, with applications to sparse linear algebra, load-balanced parallel computation, and graph computations. In recent years, graphics processing units (GPUs) have become the preferred accelerator on the world's top supercomputers. Despite this, most existing dis...
Michael S. Gilbert, E. Boman, K. Madduri· 0 citations
The results of research on parallel applications for heterogeneous architectures using the SYCL standards and the DPC++ language, as well as OpenMP, developed by students of the Department of Applied Mathematics and Artificial Intelligence (AM&AI) at MPEI during research work are presented.
As an applied problem, the s...
Olga Shamayeva, A. Chernetsov, L. Zinchenko· Russian Digital Libraries Jo...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.