Roofline Analysis of Tightly-Coupled CPU-GPU Superchips: A Study on MI300A and GH200
Open Access
- 15 November 2025
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1205-1216
Abstract
No abstract availableFunding Information
- Advanced Scientific Computing Research, Scientific Discovery through Advanced Computing (SciDAC) program, in the U.S. Department of Energy, Office of Science (DE-AC02-05CH11231)
- National Energy Research Scientific Computing Center, which is supported by the Office of Science of the U.S. Department of Energy (DE-AC02-05CH11231)
This publication has 18 references indexed in Scilit:
- Sum Reduction with OpenMP Offload on NVIDIA Grace-Hopper System2024
- Realizing the AMD Exascale Heterogeneous Processor Vision : Industry Product2024
- A comparison of two performance portability metricsConcurrency and Computation: Practice and Experience, 2023
- 9.3 NVLink-C2C: A Coherent Off Package Chip-to-Chip Interconnect with 40Gbps/pin Single-ended Signaling2023
- Performance Analysis of GPU Programming Models Using the Roofline Scaling Trajectories2020
- Performance Portability Evaluation of OpenCL Benchmarks across Intel and NVIDIA Platforms2020
- Hierarchical Roofline analysis for GPUs: Accelerating performance optimization for the NERSC‐9 Perlmutter systemConcurrency and Computation: Practice and Experience, 2019
- Implications of a metric for performance portabilityFuture Generation Computer Systems, 2019
- A quantitative roofline model for GPU kernel performance estimation using micro-benchmarks and hardware metric profilingJournal of Parallel and Distributed Computing, 2017
- Roofline Model Toolkit: A Practical Tool for Architectural and Program Analysis2015