Vau da muntanialas: Energy-efficient multi-die scalable acceleration of RNN inference
share
Recurrent neural networks such as Long Short-Term Memories (LSTMs) learn temporal dependencies by keeping an internal state, making them ideal for time-series problems such as speech recognition. However, the output-to-input feedback creates distinctive memory bandwidth and scalability challenges in designing accelerators for RNNs. We present Muntaniala, an RNN accelerator architecture for LSTM inference with a silicon-measured energy-efficiency of 3.25TOP/s/W and performance of 30.53GOP/s in UMC 65 nm technology. The scalable design of Muntaniala allows running large RNN models by combining multiple tiles in a systolic array. We keep all parameters stationary on every die in the array, drastically reducing the I/O communication to only loading new features and sharing partial results with other dies. For quantifying the overall system power, including I/O power, we built Vau da Muntanialas, to the best of our knowledge, the first demonstration of a systolic multi-chip-on-PCB array of RNN accelerator. Our multi-die prototype performs LSTM inference with 192 hidden states in 330μ s with a total system power of 9.0mW at 10MHz consuming 2.95μ J. Targeting the 8/16-bit quantization implemented in Muntaniala, we show a phoneme error rate (PER) drop of approximately 3 respect to floating-point (FP) on a 3L-384NH-123NI LSTM network on the TIMIT dataset.
READ FULL TEXT