NVIDIA Developer Blog

Estimating Visual Odometry with Prerecorded Stereo Images and the NVIDIA Isaac SDK

Fri, 22 Jan 2021 22:36:06 GMT

The NVIDIA Isaac SDK includes a production-ready, real-time visual odometry solution. According to the KITTI visual odometry algorithm comparison benchmark, it is 11th for accuracy and the fastest for performance. Figure 2 shows that Isaac provides better speed and accuracy than the widely used ORB Slam2. This post shows you how to run Isaac SDK […]html

Building the Smart Cloud Using the Best SmartNICs and DPUs, Part 1

Fri, 22 Jan 2021 22:35:51 GMT

This post was originally published on the Mellanox blog. Amazon recently announced that Alexa, the smart personal voice assistant is expanding from a small desktop gadget and getting into the fabric of our life. Anything and everything can get an Alexa boost, including your microwave, home security, car infotainment, and even your wall clock. Soon, […]html

Building the Smart Cloud Using the Best SmartNICs and DPUs, Part 2

Fri, 22 Jan 2021 22:34:54 GMT

This post was originally published on the Mellanox blog. In part one, I said that the smart devices around us are changing our lives in remarkable ways. However, the infrastructure to support these smart innovations hasn’t fully evolved in terms of flexibility, performance, and efficiency. A software-defined world offers flexibility but at the cost of […]html

Scaling Out the Deep Learning Cloud Efficiently

Fri, 22 Jan 2021 22:31:01 GMT

The Duchess of Windsor famously said that you can never be too rich or too thin. A similar observation is true when trying to match deep learning applications and compute resources: You can never have too much horsepower. Intractable problems in fields as diverse as finance, security, medical research, resource exploration, self-driving vehicles, and defense […]html

Accelerating NVSHMEM 2.0 Team-Based Collectives Using NCCL

Fri, 22 Jan 2021 21:47:32 GMT

NVSHMEM 2.0 is introducing a new API for performing collective operations based on the Team Management feature of the OpenSHMEM 1.5 specification. A team is a subset of processing elements (PEs) in an OpenSHMEM job. The concept is analogous to communicators in MPI. The new Teams API is a replacement for the active-set-based API for […]html

Accelerating Connection Tracking to Turbo-Charge Stateful Security

Thu, 21 Jan 2021 00:59:21 GMT

Public cloud and telecommunication service providers are moving in the direction of large hyperscale datacenters like Google and Amazon. They are looking for ways to increase efficiencies, flexibility, and agility, and so are turning towards disaggregation and server virtualization as critical tenets of their modernization efforts. However, in doing so, they are stumbling across several […]html

Researching and Developing an Autonomous Vehicle Lane-Following System

Wed, 20 Jan 2021 22:24:33 GMT

Four years ago, a system known as PilotNet became the first NVIDIA system to steer an autonomous car along a roadway. This system represents a departure from the classical approach for self-driving in which the process is manually decomposed into a series of modules, each performing a different task. In contrast, PilotNet is a single […]html

Detecting Out-of-Band Malware with NVIDIA BlueField DPU

Thu, 14 Jan 2021 18:22:35 GMT

In an era where cyberthreats are around every corner and with increasing attacks on data centers, security has become an essential element to include in every machine guarding user data. However, many security offerings are defenseless in the presence of malware. Furthermore, software-based security consumes compute and memory resources that should be allocated to users. […]html

Training Your JetBot in NVIDIA Isaac Sim

Sat, 19 Dec 2020 01:47:41 GMT

How do you teach your JetBot new tricks? In this post, we highlight NVIDIA Isaac Sim simulation and training capabilities by walking you through how to train the JetBot in Isaac Sim with reinforcement learning (RL) and test this trained RL model on NVIDIA Jetson Nano with the real JetBot. Goal The goal is to […]html

Making Apache Spark More Concurrent

Fri, 18 Dec 2020 23:50:31 GMT

Apache Spark provides capabilities to program entire clusters with implicit data parallelism. With Spark 3.0 and the open source RAPIDS Accelerator for Spark, these capabilities are extended to GPUs. However, prior to this work, all CUDA operations happen in the default stream, causing implicit synchronization and not taking advantage of concurrency on the GPU. In […]html

Training a Text2Sparql Model with MK-SQuIT and NeMo

Fri, 18 Dec 2020 23:00:19 GMT

Across several verticals, question answering (QA) is one of the fastest ways to deliver business value using conversational AI. Informally, QA is the task of mapping a user query to an answer from a given context. This open-ended definition is best understood through an example: Question: Where are the headquarters of NVIDIA?Context: “NVIDIA Corporation (en-VID-ee-\u0259) […]html

Deploying a Scalable Object Detection Inference Pipeline: Optimization and Deployment, Part 3

Fri, 18 Dec 2020 19:29:33 GMT

This post is the third in a series on Autonomous Driving at Scale, developed with Tata Consultancy Services (TCS). The previous posts provided a general overview of deep learning inference for object detection and covered the object detection inference process and object detection metrics. In this post, we conclude with a brief look at the […]html

Deploying a Scalable Object Detection Pipeline: The Inferencing Process, Part 2

Fri, 18 Dec 2020 19:22:56 GMT

This post is the second in a series on Autonomous Driving at Scale, developed with Tata Consultancy Services (TCS). The previous post in this series provided a general overview of the deep learning inference for object detection. In this post, we dive deep into the object detection inference process. We cover an explanation of the […]html

Minimizing Deep Learning Inference Latency with NVIDIA Multi-Instance GPU

Fri, 18 Dec 2020 18:40:47 GMT

Recently, NVIDIA unveiled the A100 GPU model, based on the NVIDIA Ampere architecture. Ampere introduced many features, including Multi-Instance GPU (MIG), that play a special role for deep learning-based (DL) applications. MIG makes it possible to use a single A100 GPU as if it were multiple smaller GPUs, maximizing utilization for DL workloads and providing […]html

Optimizing Data Transfer Using Lossless Compression with NVIDIA nvcomp

Fri, 18 Dec 2020 18:38:58 GMT

One of the most interesting applications of compression is optimizing communications in GPU applications. GPUs are getting faster every year. For some apps, transfer rates for getting data in or out of the GPU can’t keep up with the increase in GPU memory bandwidth or computational power. Often, the bottleneck is the interconnect between GPUs […]html

Deploying AI Deep Learning Models with NVIDIA Triton Inference Server

Fri, 18 Dec 2020 03:30:10 GMT

In the world of machine learning, models are trained using existing data sets and then deployed to do inference on new data. In a previous post, Simplifying and Scaling Inference Serving with NVIDIA Triton 2.3, we discussed inference workflow and the need for an efficient inference serving solution. In that post, we introduced Triton Inference […]html

Streaming Simulation and Training Applications with Project Anywhere

Fri, 18 Dec 2020 02:30:07 GMT

Imagine a future where ultra-high-fidelity simulation and training applications are deployed over any network topology from a centralized secure cloud or on-premises infrastructure. Imagine that you can stream graphical training content from the datacenter to remote end devices ranging from a single flat-screen or synchronized displays to AR/VR/MR head-mounted displays. This datacenter architecture enables the […]html

Enhancing Memory Allocation with New NVIDIA CUDA 11.2 Features

Wed, 16 Dec 2020 17:51:50 GMT

CUDA is the software development platform for building GPU-accelerated applications, providing all the components needed to develop applications targeting every NVIDIA GPU platform for general purpose compute acceleration. The latest CUDA release, CUDA 11.2, is focused on improving the user experience and application performance for CUDA developers. CUDA 11.2 has several important features including programming […]html

Monitoring High-Performance Machine Learning Models with RAPIDS and whylogs

Mon, 14 Dec 2020 23:54:44 GMT

Machine learning (ML) data is big and messy. Organizations have increasingly adopted RAPIDS and cuML to help their teams run experiments faster and achieve better model performance on larger datasets. That, in turn, accelerates the training of ML models using GPUs. With RAPIDS, data scientists can now train models 100X faster and more frequently. Like […]html

Supercharging the World’s Fastest AI Supercomputing Platform on NVIDIA HGX A100 80GB GPUs

Thu, 10 Dec 2020 17:07:15 GMT

Exploding model sizes in deep learning and AI, complex simulations in high-performance computing (HPC), and massive datasets in data analytics all continue to demand faster and more advanced GPUs and platforms. At SC20, we announced the NVIDIA A100 80GB GPU, the latest addition to the NVIDIA Ampere family, to help developers, researchers, and scientists tackle […]html