FORMS: Fine-grained Polarized ReRAM-based In-situ Computation for Mixed-signal DNN Accelerator
arXiv:2106.09144
Abstract
Recent works demonstrated the promise of using resistive random access memory (ReRAM) as an emerging technology to perform inherently parallel analog domain in-situ matrix-vector multiplication -- the intensive and key computation in DNNs. With weights stored in the ReRAM crossbar cells as conductance, when the input vector is applied to word lines, the matrix-vector multiplication results can be generated as the current in bit lines. A key problem is that the weight can be either positive or negative, but the in-situ computation assumes all cells on each crossbar column with the same sign. The current architectures either use two ReRAM crossbars for positive and negative weights, or add an offset to weights so that all values become positive. Neither solution is ideal: they either double the cost of crossbars, or incur extra offset circuity. To better solve this problem, this paper proposes FORMS, a fine-grained ReRAM-based DNN accelerator with polarized weights. Instead of trying to represent the positive/negative weights, our key design principle is to enforce exactly what is assumed in the in-situ computation -- ensuring that all weights in the same column of a crossbar have the same sign. It naturally avoids the cost of an additional crossbar. Such weights can be nicely generated using alternating direction method of multipliers (ADMM) regularized optimization, which can exactly enforce certain patterns in DNN weights. To achieve high accuracy, we propose to use fine-grained sub-array columns, which provide a unique opportunity for input zero-skipping, significantly avoiding unnecessary computations. It also makes the hardware much easier to implement. Putting all together, with the same optimized models, FORMS achieves significant throughput improvement and speed up in frame per second over ISAAC with similar area cost.
In Proceedings of the 48th Annual International Symposium on Computer Architecture (ISCA), 2021
References in corpus (10)
- Adam: A Method for Stochastic Optimization
- AMC: AutoML for Model Compression and Acceleration on Mobile Devices
- A Systematic DNN Weight Pruning Framework using Alternating Direction Method of Multipliers
- Basic Performance Measurements of the Intel Optane DC Persistent Memory Module
- Multiscale Co-Design Analysis of Energy, Latency, Area, and Accuracy of a ReRAM Analog Neural Training Accelerator
- RAPIDNN: In-Memory Deep Neural Network Acceleration Framework
- YOLObile: Real-Time Object Detection on Mobile Devices via Compression-Compilation Co-Design
- SS-Auto: A Single-Shot, Automatic Structured Weight Pruning Framework of DNNs with Ultra-High Efficiency
- A SOT-MRAM-based Processing-In-Memory Engine for Highly Compressed DNN Implementation
- TIMELY: Pushing Data Movements and Interfaces in PIM Accelerators Towards Local and in Time Domain