1.0 Introduction

Loading...

En provenance du cours de Intel

Fundamentals of Parallelism on Intel Architecture

62 notes

Intel

Fundamentals of Parallelism on Intel Architecture

62 notes

This course will introduce you to the multiple forms of parallelism found in modern Intel architecture processors and teach you the programming frameworks for handling this parallelism in applications. You will get access to a cluster of modern manycore processors (Intel Xeon Phi architecture) for experiments with graded programming exercises. This course can apply to various HPC and datacenter workloads and framework including artificial intelligence (AI). You will learn how to handle data parallelism with vector instructions, task parallelism in shared memory with threads, parallelism in distributed memory with message passing, and memory architecture parallelism with optimized data containers. This knowledge will help you to accelerate computational applications by orders of magnitude, all the while keeping your code portable and future-proof. Prerequisite: programming in C/C++ or Fortran in the Linux environment and Linux shell proficiency (navigation, file copying, editing files in text-based editors, compilation).

À partir de la leçon

Modern Code

In the Introduction we will learn...

1.0 Introduction3:51

1.1 Why this course?5:34

Rencontrer les enseignants

Andrey Vladimirov
Head of High-Performance Computing Research
Colfax International

Hello, everyone, and

welcome to Fundamentals of Parallelism on Intel Architecture.

Exactly 20 years ago, this supercomputer was being built at the Sandia Labs.

It was ASCI Red, the world's fastest supercomputer of the day.

These four rows of cabinets contain over 9,000 Intel Pentium Pro Processors.

Together, they deliver over two trillion floating point operations per second.

Two trillion, with a t, floating point operations per second.

Just imagine the science you can do on this thing.

Predict weather, develop life saving drugs, understand protein folding,

play chess better than the best human player, or

simulate the conditions of the big bang.

Today, in 2017, we can get the same performance

in a slightly smaller package, slightly smaller, about this big.

This is an Intel Xeon Phi Processor.

It can perform over 2 trillion floating point operations in double precision per

second sitting quietly under your desk.

The complexity of what used to be the world's fastest supercomputer

has been compressed down to a single chip.

So now vast numbers of people can have access to this computing power or

to the computing power of large installations of such processors.

This introduces a new question.

How many people know how to program such computers?

I come from the physics background.

And in my life as physicist, I develop computer models

to simulate how cosmic rays are produced and what impact they have on our universe.

In my interactions with fellow researchers in natural sciences

I noticed a common sentiment.

People think why does computer programming have to be so hard?

Do I have to become a computer scientist to learn how to use my computer and

do physics?

And I can absolutely relate to that sentiment.

Computers are evolving faster than the software and programming methods for them.

And it's not making things easier that educational programs are often

lagging behind the software and the best known methods.

So engineers and

scientists are not opposed to learning, they are opposed to the unknown.

>> After spending so many hours on learning,

do they come out with something that's useless?

Do they come out with something that's transient?

From my experience with computer science, I know that, indeed,

programming modern processors is very hard.

But at the same time, for people not familiar with parallel programming,

there is a lot of low hanging fruit,

something that they can learn quickly to reap the benefits of parallelism and

accelerate their code by one or two orders of magnitude.

So I'm really excited to be creating these courses.

We decided to separate the learning process into two parts.

In the first part, fundamentals of parallelism.

People not familiar with parallel programming will quickly learn

what they can do to make their applications run faster.

And then, in the second part, performance optimization.

Those who wish to dig deeper will get a lot of additional information.

High performance computing enables exciting science.

Finding cures for diseases and predicting natural disasters.

The development of clean energy and the protection of the environment.

It helps to make automobiles safer, teach computers to detect crime, and

create machines that help people with disabilities.

Thank you for learning with us.

I hope that you can take the tools that we give you and apply them to your field,

make your applications run faster and, through that, fuel discovery.

Enjoy the course, good luck.

Explorer notre catalogue

Rejoignez-nous gratuitement et obtenez des recommendations, des mises à jour et des offres personnalisées.

Coursera propose un accès universel à la meilleure formation au monde, en partenariat avec des universités et des organisations du plus haut niveau, pour proposer des cours en ligne.

© 2019 Coursera Inc. Tous droits réservés.

Télécharger dans l'App Store

Disponible sur Google Play