Inter-APU Communication on AMD MI300A Systems via Infinity Fabric: a Deep Dive
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909739445649408 |
|---|---|
| author | Schieffer, Gabin Wahlgren, Jacob Shi, Ruimin León, Edgar A. Pearce, Roger Gokhale, Maya Peng, Ivy |
| author_facet | Schieffer, Gabin Wahlgren, Jacob Shi, Ruimin León, Edgar A. Pearce, Roger Gokhale, Maya Peng, Ivy |
| contents | The ever-increasing compute performance of GPU accelerators drives up the need for efficient data movements within HPC applications to sustain performance. Proposed as a solution to alleviate CPU-GPU data movement, AMD MI300A Accelerated Processing Unit (APU) combines CPU, GPU, and high-bandwidth memory (HBM) within a single physical package. Leadership supercomputers, such as El Capitan, group four APUs within a single compute node, using Infinity Fabric Interconnect. In this work, we design specific benchmarks to evaluate direct memory access from the GPU, explicit inter-APU data movement, and collective multi-APU communication. We also compare the efficiency of HIP APIs, MPI routines, and the GPU-specialized RCCL library. Our results highlight key design choices for optimizing inter-APU communication on multi-APU AMD MI300A systems with Infinity Fabric, including programming interfaces, allocators, and data movement. Finally, we optimize two real HPC applications, Quicksilver and CloverLeaf, and evaluate them on a four MI100A APU system. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_11298 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Inter-APU Communication on AMD MI300A Systems via Infinity Fabric: a Deep Dive Schieffer, Gabin Wahlgren, Jacob Shi, Ruimin León, Edgar A. Pearce, Roger Gokhale, Maya Peng, Ivy Distributed, Parallel, and Cluster Computing The ever-increasing compute performance of GPU accelerators drives up the need for efficient data movements within HPC applications to sustain performance. Proposed as a solution to alleviate CPU-GPU data movement, AMD MI300A Accelerated Processing Unit (APU) combines CPU, GPU, and high-bandwidth memory (HBM) within a single physical package. Leadership supercomputers, such as El Capitan, group four APUs within a single compute node, using Infinity Fabric Interconnect. In this work, we design specific benchmarks to evaluate direct memory access from the GPU, explicit inter-APU data movement, and collective multi-APU communication. We also compare the efficiency of HIP APIs, MPI routines, and the GPU-specialized RCCL library. Our results highlight key design choices for optimizing inter-APU communication on multi-APU AMD MI300A systems with Infinity Fabric, including programming interfaces, allocators, and data movement. Finally, we optimize two real HPC applications, Quicksilver and CloverLeaf, and evaluate them on a four MI100A APU system. |
| title | Inter-APU Communication on AMD MI300A Systems via Infinity Fabric: a Deep Dive |
| topic | Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2508.11298 |