Speed of Light Exact Greedy Decoding for RNN-T Speech Recognition Models on GPU

TL;DR

This paper introduces a GPU-based exact greedy decoding method for RNN-T speech recognition models that significantly reduces inference time and GPU idle periods, enabling high-throughput performance for large models.

Contribution

The authors present a novel CUDA-based implementation of greedy decoding that eliminates GPU idle time, achieving substantial speedups for large RNN-T models.

Findings

Speedup of 2.5x for a 1.1B parameter RNN-T model

Achieved near-CTC model inference speed for large RNN-T models

Applicable to label looping greedy decoding with notable speed improvements

Abstract

The vast majority of inference time for RNN Transducer (RNN-T) models today is spent on decoding. Current state-of-the-art RNN-T decoding implementations leave the GPU idle ~80% of the time. Leveraging a new CUDA 12.4 feature, CUDA graph conditional nodes, we present an exact GPU-based implementation of greedy decoding for RNN-T models that eliminates this idle time. Our optimizations speed up a 1.1 billion parameter RNN-T model end-to-end by a factor of 2.5x. This technique can applied to the "label looping" alternative greedy decoding algorithm as well, achieving 1.7x and 1.4x end-to-end speedups when applied to 1.1 billion parameter RNN-T and Token and Duration Transducer models respectively. This work enables a 1.1 billion parameter RNN-T model to run only 16% slower than a similarly sized CTC model, contradicting the common belief that RNN-T models are not suitable for high throughput inference. The implementation is available in NVIDIA NeMo.