Multi-objective hardware-mapping co-optimisation for multi-DNN workloads on chiplet-based accelerators
The need to efficiently execute different Deep Neural Networks (DNNs) on the same computing platform, coupled with the requirement for easy scalability, makes Multi-Chip Module (MCM)-based accelerators a preferred design choice. Such an accelerator brings together heterogeneous sub-accelerators in t...
| Autores: | , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2024 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/410277 |
| Acceso en línea: | https://hdl.handle.net/2117/410277 https://dx.doi.org/10.1109/TC.2024.3386067 |
| Access Level: | acceso abierto |
| Palabra clave: | Neural networks (Computer science) Energy consumption Deep Neural Network (DNN) Accelerator Design Space Exploration (DSE) Hardware-mapping co-optimisation Xarxes neuronals (Informàtica) Energia -- Consum Àrees temàtiques de la UPC::Informàtica::Arquitectura de computadors |
| Sumario: | The need to efficiently execute different Deep Neural Networks (DNNs) on the same computing platform, coupled with the requirement for easy scalability, makes Multi-Chip Module (MCM)-based accelerators a preferred design choice. Such an accelerator brings together heterogeneous sub-accelerators in the form of chiplets, interconnected by a Network-on-Package (NoP). This paper addresses the challenge of selecting the most suitable sub-accelerators, configuring them, determining their optimal placement in the NoP, and mapping the layers of a predetermined set of DNNs spatially and temporally. The objective is to minimise execution time and energy consumption during parallel execution while also minimising the overall cost, specifically the silicon area, of the accelerator. This paper presents MOHaM, a framework for multi-objective hardware-mapping co-optimisation for multi-DNN workloads on chiplet-based accelerators. MOHaM exploits a multi-objective evolutionary algorithm that has been specialised for the given problem by incorporating several customised genetic operators. MOHaM is evaluated against state-of-the-art Design Space Exploration (DSE) frameworks on different multi-DNN workload scenarios. The solutions discovered by MOHaM are Pareto optimal compared to those by the state-of-the-art. Specifically, MOHaM-generated accelerator designs can reduce latency by up to 96% and energy by up to 96.12%. |
|---|