Demystifying Map Space Exploration for NPUs

10/07/2022
by   Sheng-Chun Kao, et al.
0

Map Space Exploration is the problem of finding optimized mappings of a Deep Neural Network (DNN) model on an accelerator. It is known to be extremely computationally expensive, and there has been active research looking at both heuristics and learning-based methods to make the problem computationally tractable. However, while there are dozens of mappers out there (all empirically claiming to find better mappings than others), the research community lacks systematic insights on how different search techniques navigate the map-space and how different mapping axes contribute to the accelerator's performance and efficiency. Such insights are crucial to developing mapping frameworks for emerging DNNs that are increasingly irregular (due to neural architecture search) and sparse, making the corresponding map spaces much more complex. In this work, rather than proposing yet another mapper, we do a first-of-its-kind apples-to-apples comparison of search techniques leveraged by different mappers. Next, we extract the learnings from our study and propose two new techniques that can augment existing mappers – warm-start and sparsity-aware – that demonstrate speedups, scalability, and robustness across diverse DNN models.

READ FULL TEXT

page 2

page 9

research
05/27/2021

NAAS: Neural Accelerator Architecture Search

Data-driven, automatic design space exploration of neural accelerator ar...
research
04/26/2021

HAO: Hardware-aware neural Architecture Optimization for Efficient Inference

Automatic algorithm-hardware co-design for DNN has shown great success i...
research
02/18/2020

Marvel: A Data-centric Compiler for DNN Operators on Spatial Accelerators

The efficiency of a spatial DNN accelerator depends heavily on the compi...
research
03/02/2021

Mind Mappings: Enabling Efficient Algorithm-Accelerator Mapping Space Search

Modern day computing increasingly relies on specialization to satiate gr...
research
11/04/2021

RT-RCG: Neural Network and Accelerator Search Towards Effective and Real-time ECG Reconstruction from Intracardiac Electrograms

There exists a gap in terms of the signals provided by pacemakers (i.e.,...
research
01/23/2023

Enabling Hard Constraints in Differentiable Neural Network and Accelerator Co-Exploration

Co-exploration of an optimal neural architecture and its hardware accele...
research
02/18/2020

MARVEL: A Decoupled Model-driven Approach for Efficiently Mapping Convolutions on Spatial DNN Accelerators

The efficiency of a spatial DNN accelerator depends heavily on the compi...

Please sign up or login with your details

Forgot password? Click here to reset