[Paid Event] Nvidia CUDA 10 & Tensorflow 1.11 -- Hands-on Workshop
dim. 11 novembre à 20:00
We will be discussing how to compile and install from source a GPU accelerated instance of Tensorflow in Ubuntu 18.04 LTS. Tensorflow is a deep-learning framework developed by Google. It has become an industry standard tool for both deep-learning research and production grade application development. Ubuntu[masked] LTS (Bionic Beaver) is the latest long term support variant of Ubuntu linux. It will be supported for 5 years until April 2023. It is one of the favorite choice of linux distribution for deploying scalable deep-learning applications, both in research and in production settings. Read more about this dsitribution here: https://wiki.ubuntu.com/BionicBeaver/ReleaseNotes This session will cover the basics of Nvidia CUDA and use CUDA to accelerate Tensorflow applications. Nvidia CUDA is a parallel computing platform and programming model for general computing on graphical processing units (GPUs) from Nvidia. CUDA handles the GPU acceleration of deep-learning tasks using Tensorflow. Released on September 19, 2018, Nvidia CUDA 10 is the latest release of CUDA. We will be using CUDA 10, Tnesorflow 1.11 and TensorRT 5.0 for the workshop. We will also discuss Nvidia CUDA Deep Neural Network library (cuDNN), a GPU-accelerated library of primitives for deep neural networks. cuDNN provides highly tuned implementations for standard routines such as forward and backward convolution, pooling, normalization, and activation layers. cuDNN is part of the Nvidia Deep Learning SDK. Another key piece of technology for GPU acceleration using CUDA is the NVIDIA Collective Communications Library (NCCL), which implements multi-GPU and multi-node collective communication primitives that are performance optimized for NVIDIA GPUs. NCCL provides routines such as all-gather, all-reduce, broadcast, reduce, reduce-scatter, that are optimized to achieve high bandwidth over PCIe and NVLink high-speed interconnect. Developers of deep learning frameworks and HPC applications can rely on NCCL’s highly optimized, MPI compatible and topology aware routines, to take full advantage of all available GPUs within and across multiple nodes. This allows them to focus on developing new algorithms and software capabilities, rather than performance tuning low-level communication collectives. Tensorflow uses NCCL to deliver near-linear scaling of deep learning training on multi-GPU systems. This is a paid workshop. Access to the GPU powered virtual machines (VMs) for hands-on deployments are only available for participants who purchased a ticket. Last minute purchase of tickets may not receive a VM due to allocation bottlenecks. Remember to buy the tickets at-least 24 hours before the event. The tickets are available for purchase here: https://www.moad.computer/store/p36/Healthcare_Analytics The goal of this workshop is to learn how to leverage GPU to accelerate applications such as Tensorflow. Please bring a laptop to follow along the content effectively. The content of this session is loosely based on this blog post: http://remananr.com/Blog/tensorflow_ubuntu_1804lts/ Requirements:
1) Basics of shell scripting
2) Basics of python 3
3) Familiarity with tools like nano and screen in linux TLDR:
Tensorflow 1.11 with CUDA 10 and TensorRT 5.0
Nous avons temporairement désactivé la possibilité de naviguer vers les tags.