MEMTI: Optimizing On-Chip Nonvolatile Storage for Visual Multitask Inference at the Edge
- 4 October 2019
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Micro
- Vol. 39 (6) , 73-81
Abstract
The combination of specialized hardware and embedded non-volatile memories (eNVM) holds promise for energy-efficient DNN inference at the edge. However, integrating DNN hardware accelerators with eNVMs still presents several challenges. Multi-level programming is desirable for achieving maximal storage density on chip, but the stochastic nature of eNVM writes makes them prone to errors and further increases the write energy and latency. We present MEMTI, a memory architecture that leverages a multi-task learning technique for maximal reuse of DNN parameters across multiple visual tasks. We show that by retraining and updating only 10% of all DNN parameters, we can achieve efficient model adaptation across a variety of visual inference tasks. The system performance is evaluated by integrating the memory with the open-source NVIDIA Deep Learning Architecture (NVDLA).Keywords
Funding Information
- Applications Driving Architectures (ADA) Research Center
This publication has 5 references indexed in Scilit:
- Efficient Parametrization of Multi-domain Deep Neural Networks2018
- Through the Looking Glass - The 2018 Edition: Trends in Solid-State Circuits from the 65th ISSCCIEEE Solid-State Circuits Magazine, 2018
- Deep Learning for Computer ArchitectsSynthesis Lectures on Computer Architecture, 2017
- The Impact of Self-Heating on Charge Trapping in High- $k$ -Metal-Gate nFETsIEEE Electron Device Letters, 2015
- NVSim: A Circuit-Level Performance, Energy, and Area Model for Emerging Nonvolatile MemoryIEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2012