High-Performance Deep Learning via a Single Building Block
arXiv:1906.06440
Abstract
Deep learning (DL) is one of the most prominent branches of machine learning. Due to the immense computational cost of DL workloads, industry and academia have developed DL libraries with highly-specialized kernels for each workload/architecture, leading to numerous, complex code-bases that strive for performance, yet they are hard to maintain and do not generalize. In this work, we introduce the batch-reduce GEMM kernel and show how the most popular DL algorithms can be formulated with this kernel as the basic building-block. Consequently, the DL library-development degenerates to mere (potentially automatic) tuning of loops around this sole optimized kernel. By exploiting our new kernel we implement Recurrent Neural Networks, Convolution Neural Networks and Multilayer Perceptron training and inference primitives in just 3K lines of high-level code. Our primitives outperform vendor-optimized libraries on multi-node CPU clusters, and we also provide proof-of-concept CNN kernels targeting GPUs. Finally, we demonstrate that the batch-reduce GEMM kernel within a tensor compiler yields high-performance CNN primitives, further amplifying the viability of our approach.
References in corpus (7)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
- cuDNN: Efficient Primitives for Deep Learning
- Low-memory GEMM-based convolution algorithms for deep neural networks
- Stripe: Tensor Compilation via the Nested Polyhedral Model
- On Scale-out Deep Learning Training for Cloud and HPC
- Large-Batch Training for LSTM and Beyond