Chipmunk: A Systolically Scalable 0.9 mm${}^2$, 3.08 Gop/s/mW @ 1.2 mW Accelerator for Near-Sensor Recurrent Neural Network Inference
This enables zero-latency voice-based human-machine interfaces on low-power devices, representing an incremental improvement in hardware efficiency for specific RNN applications.
The paper tackles the problem of on-device recurrent neural network inference for low-power mobile and wearable devices by presenting Chipmunk, a small hardware accelerator that achieves a measured peak efficiency of 3.08 Gop/s/mW at 1.24 mW peak power and consumes less than 13 mW average power for real-time phoneme extraction.
Recurrent neural networks (RNNs) are state-of-the-art in voice awareness/understanding and speech recognition. On-device computation of RNNs on low-power mobile and wearable devices would be key to applications such as zero-latency voice-based human-machine interfaces. Here we present Chipmunk, a small (<1 mm${}^2$) hardware accelerator for Long-Short Term Memory RNNs in UMC 65 nm technology capable to operate at a measured peak efficiency up to 3.08 Gop/s/mW at 1.24 mW peak power. To implement big RNN models without incurring in huge memory transfer overhead, multiple Chipmunk engines can cooperate to form a single systolic array. In this way, the Chipmunk architecture in a 75 tiles configuration can achieve real-time phoneme extraction on a demanding RNN topology proposed by Graves et al., consuming less than 13 mW of average power.