Xavier/Deep Learning/Deep Learning Accelerator

NVIDIA®Jetson Xavier™
Introduction
SoM Overview Carrier Boards NVIDIA Xavier Developer Kit Connect Tech Rogue Connect Tech Mimic Software Support NVIDIA JetPack
JetPack 5.X
Getting Started Installing JetPack Flashing the board with GUI installer JetPack Components Compiling Source Code Flashing the Board from Cmdline Performance Tuning Evaluating Performance Set Values Manually Tuning Power Maximizing Performance
Development in the Board
Getting into the Board Using Serial Console Using SSH Using Ubuntu Login Installing Packages
Video Capture and Display
Description Supported Cameras Imx219 Driver Software Support Using GStreamer Using Libargus
Processors
HD Audio and Video Subystem Video Encoder Video Decoder JPEG Encoder and Decoder Video Image Compositor Audio Processing Engine Deep Learning Accelerator Volta GPU Description Using CUDA Using OpenGL
Interfaces
I2C SD/eMMC Controller USB Controller
GStreamer Pipelines
Capture and Display H264 H265 VP9 MPEG4 Simultaneous Streams Compositor
Deep Learning
TensorRT Using TensorRT integrated with Tensorflow Parsing a Tensorflow model for TensorRT Parsing Caffe model for TensorRT Building TensorRT API examples Deep Learning Accelerator Software Support Deep Learning Tutorials Jetson Inference Jetson Reinforcement
RidgeRun Products
GstPTZR GstCUDA GstColorTransfer GstWebRTC GstRtspSink GstInterpipe GstInference GStreamer Daemon GstShark
Reference Documentation
Contact Us

The NVIDIA Deep Learning Accelerator (NVDLA) is a fixed function engine used to accelerate inference operations on convolution neural networks (CNNs). You will learn about the software available to work with the Deep Learning Accelerator and how to use it to create your own projects. If you want to learn more about this engine, its hardware, and the open-source project NVDLA please refer to the NVDLA Processor wiki page.

NVDLA software design is grouped into two groups: the compilation tools (model conversion), and the runtime environment (run-time software to load and execute networks on NVDLA). The general flow of this is as shown in the figure below, and each of these is described below.

Front view of NVIDIA Jetson Xavier developer kit

Compilation Tools

Compilation tools include a compiler, parser, and optimizer.

• Parser – Reads a pre-trained Caffe model and an intermediate representation to pass to the compiler.
• Compiler – creates a sequence of hardware layers optimized for an NVDLA configuration.

These steps can be performed offline on the host device and then deploy the loadable model on Xavier's NVDLAs. Knowing the specific NVDLA hardware configuration enables the compiler to further optimize inference. However, the specific NVDLA configuration on the Tegra Xavier is not yet available.

Runtime Environment

The runtime environment involves running a model on compatible NVDLA hardware. It is effectively divided into two layers:

• User Mode Driver – The main interface with user-mode programs. After parsing the neural network compiler compiles the network layer by layer and converts it into a file format called NVDLA Loadable. User mode runtime driver loads this loadable and submits inference job to Kernel Mode Driver
• Kernel Mode Driver – Consists of drivers and firmware that do the work of scheduling layer operations on NVDLA and programming the NVDLA registers to configure each functional block.

Selecting which DLA to use

You can select which of the two DLAs to use with IBuilder::setDLACore() and IRuntime::setDLACore()