Introduction to NVIDIA DALI: GPU-accelerated Data Preprocessing
, Senior Software Development Engineer, NVIDIA
As GPUs’ computational power increases, providing deep-learning models with data as fast as it's consumed gets more difficult. Data preprocessing, typically executed on the CPU, becomes the bottleneck of the system. A solution to this problem is to offload data preprocessing to the GPU. NVIDIA DALI offers tools for creating GPU-accelerated data pipelines for deep-learning applications, supporting a wide range of generic and data domain-specific data processing operations. DALI can be easily integrated with your deep-learning framework of choice, and can also be deployed at inference time, thanks to the Triton Inference Server support. We'll walk you through the basic concepts of DALI. You'll learn how to write a DALI pipeline, and how to integrate into your training or inference application. Be sure to also tune in to "Effective NVIDIA DALI: Accelerating Real-life Deep-learning Applications," where we'll cover practical aspects of using DALI in a hands-on fashion.