Close

Presentation

SparkXD: A Framework for Resilient and Energy-Efficient Spiking Neural Network Inference using Approximate DRAM
Time
Location
Event Type
Research Manuscript
Virtual Programs
Hosted in Virtual Platform
Keywords
Emerging Models of Computation
Topics
Design
DescriptionThe DRAM accesses dominate the energy consumption in processing spiking neural networks (SNNs). State-of-the-art in SNN-systems do not optimize the DRAM energy-per-access, thereby hindering achieving high energy-efficiency. To minimize the DRAM energy, a key knob is to reduce the DRAM supply voltage, but this may lead to DRAM errors (i.e., the so-called approximate DRAM). Towards this, we propose SparkXD, a novel framework that employs 1) error-aware SNN training; 2) SNN error-tolerance analysis to find the tolerable error rates; and 3) error-aware DRAM mapping. For 1% tolerable accuracy loss, SparkXD reduces the DRAM energy by 39.46% on average.