Dell Science Workstation Installation Manual

Dell Precision Data

Science Workstation

Guided Install Edition

Installation Manual

Release 2.0

Introduction

Albert Einstein is widely credited as saying, “The definition of insanity is doing the same thing over and over again, but expecting different results.” Converse to that, the beauty of machine learning is that it allows hardware to learn from previous computations to produce reliable, repeatable decisions and results. But of what use is that, if the process still has to rely on human guidance and intervention?

To turn potential into reality, Dell has developed Data Science Workstations (DSWs), a curated set of Precision workstations pre-bundled with hardware and software, tailored for Machine Learning and Deep Learning developers.

Those in the know are fully aware that for Machine Learning to truly be successful and barriers removed, developers

need access to the highest levels of computation power. Dell’s Precision workstations deliver the power to deploy and manage cognitive technology platforms, including Machine Learning

(ML), Artificial Intelligence (AI) and Deep

Learning (DL).

Therefore, Dell’s DSW systems are built

with the latest NVIDIA® Quadro RTX™ GPUs (RTX™ 8000/6000/5000), Intel® Xeon® CPUs, ECC memory, and advanced storage solutions, all available on the Dell website.

Backed by Dell’s turnkey approach to IT, and allowing developers to hit the ground running, these solutions are certified and listed as supported “NVIDIA® NGC-Ready” platforms on the NVIDIA® site.

NGC-Ready system validation includes tests of:

•Single and multi-GPU Deep Learning training using TensorFlow, PyTorch and NVIDIA® DeepStream Transfer Learning Toolkit

•High volume, low latency inference using NVIDIA® TensorRT, TensorRT Inference Server, and DeepStream

•Data Science ML using RAPIDS and XGBoost

•Application development using the CUDA Toolkit

In fact, these GPU-accelerated DSWs are designed to get the developer working on AI/ML/DL datasets the same day the DSW arrives from the Dell factory. With that on the table, the power is in the developer’s hands to experiment, prepare

data, and train AI/ML/DL models orders of magnitude faster than possible on CPUonly AI development platforms. It couldn’t be simpler.

Theory is one thing, practice is another, so Dell has created this step-by-step installation manual, designed to help customers easily install the “NVIDIA® Data Science Software, powered by RAPIDS” onto the DSW received from the Dell factory pre-loaded with a custom Ubuntu 18.04 LTS image. A manual that will also prove incredibly useful in instructing users

on how to install the NVIDIA® Data Science Software onto a user-installed Ubuntu 18.04 LTS OEM image provided by Dell.

The intent of this short guide is to help users quickly set up the OS and the NVIDIA® Data Science Software specifically on Dell DSWs. It is also intended to help them overcome the three time-consuming hurdles to getting started as a data science developer. These are:

For best practices, and step-by-step guidance on installing Ubuntu OS on your Data Science Workstation, visit our appendix here

To get developers started using their Data Science Workstation as quickly as possible, we offer NVIDIA® NGC-Ready Dell Precision workstation configurations validated for the

NVIDIA® Data Science Software.

1.Configuring a system with the right hardware components.

2.Installing an OS that assures support for all devices out-of-the box.

3.Compiling a curated software stack with all the necessary GPU-accelerated libraries and machine learning and deep learning frameworks.

<![if ! IE]>

<![endif]>Contents

Introduction

Pages 2-3

The way forward

– Configurations for ML/DL

Pages 6-7

Getting things started

– OS image and installation

Pages 8-9

Powering up your performance

– Setup and optimizations

Pages 10-11

Tool up

– Installing the NVIDIA® Data Science Stack v2.3.x

Pages 12-13

Appendix

– Ubuntu Installation and

Re-installation Guide

Pages 14-16

Dell Science Workstation Installation Manual

1. The way forward –

Configurations for ML/DL

Power is nothing without control, or direction. As such, Table 1 below includes recommended configurations of mobile, tower and rack Data Science Workstations that are suitable for ML/DL.

The list ranges from mobile workstations and mid-sized towers to larger tower and rack workstations with multiple GPUs & CPUs.

These configurations are suitable for data preparation, all types and sizes of ML/

DL model training, and running the resulting inference engines. But that’s not the end. Find more configuration information by visiting the Dell Precision Workstation Solutions webpage.

Precisely developed for your needs

Precision 7550:

Precision 7750:

Precision 5820 Tower:

Dell’s most powerful 15”

The most powerful Precision

Ideal for cognitive solution

mobile workstation is AI ready,

development and inference

featuring the NVIDIA® Quadro

applications. Features

RTX™ 5000 graphic card.

NVIDIA® Quadro RTX™ 6000

or 8000 graphic cards.

Precision 7920 Tower and Rack:

Handles learning model training and larger solution frameworks with ease. Features NVIDIA® Quadro RTX™ 6000 or 8000 graphic cards. And give up no performance when you remote log into the 7920 Rack from home.

+ 7 hidden pages