An MRAM-Based Deep In-Memory Architecture for Deep Neural Networks
| Autor: | Naresh R. Shanbhag, Ameya D. Patil, Sujan K. Gonugondla, Haocheng Hua, Mingu Kang |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
010302 applied physics
Magnetoresistive random-access memory Random access memory Artificial neural network Computer science Process (computing) 02 engineering and technology 01 natural sciences Matrix multiplication 020202 computer hardware & architecture Computational science 0103 physical sciences Memory architecture 0202 electrical engineering electronic engineering information engineering Deep neural networks MNIST database |
| Zdroj: | ISCAS |
| DOI: | 10.1109/iscas.2019.8702206 |
| Popis: | This paper presents an MRAM-based deep in-memory architecture (MRAM-DIMA) to efficiently implement multi-bit matrix vector multiplication for deep neural networks using a standard MRAM bitcell array. The MRAM-DIMA achieves an 4.5 × and 70× lower energy and delay, respectively, compared to a conventional digital MRAM architecture. Behavioral models are developed to estimate the impact of circuit non-idealities, including process variations, on the DNN accuracy. An accuracy drop of ≤ 0.5% (≤ 1%) is observed for LeNet-300-100 on the MNIST dataset (a 9-layer CNN on the CIFAR-10 dataset), while tolerating 24% (12%) variation in cell conductance in a commercial 22 nm CMOS-MRAM process. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|