The Torch By ArrayFire: Q3 GPU Updates

John Melonakos ArrayFire, Newsletter Leave a Comment

News for the accelerated computing community – June 27, 2022

Signup for Newsletter Emails


Dear Friends,

Welcome to the first newsletter for our ArrayFire community!

This newsletter brings together people using and developing ArrayFire and other accelerated computing tools. You are part of this vibrant group that “gathers” together around open source work, including:

  • Domain professionals – people who are experts in their disciplines doing serious heavy-lifting from research to production
  • Machine learning gurus – people who care about the best performance for training and inference
  • Hackers – people who love to contribute highly-optimized down-to-the-metal code

You are distinguished professionals in your domains, and we hope to build more opportunities for you to interact with the ArrayFire team and one another. We will start with this lightweight quarterly newsletter.

At a glance, you’ll be able to see recent developments as well as upcoming opportunities.

Enjoy!

-John Melonakos, CEO & Co-Founder


Product Releases

ArrayFire v3.8.2 was released on May 19, 2022. Read more about it on the ArrayFire blog.

Project Spotlight:  Flashlight by Meta

Flashlight is a fast, flexible machine learning library written entirely in C++ from Facebook AI Research and the creators of Torch, TensorFlow, Eigen, and Deep Speech. Its core features include:

  • Total internal modifiability, including internal APIs for tensor computation
  • A small footprint, with the core clocking in at under 10 MB and 20k lines of C++
  • High-performance defaults featuring just-in-time kernel compilation with modern C++ via the ArrayFire tensor library
  • An emphasis on efficiency and scale

Native support in C++ and simple extensibility make Flashlight a powerful research framework that’s *hackable to its core* and enables fast iteration on new experimental setups and algorithms without sacrificing performance.

In a single repository, Flashlight provides apps for research across multiple domains:

Flashlight is fast, beating TensorFlow and PyTorch in 11 of 12 benchmarks (results available on arXiv).

Table 3

User Showcase

ArrayFire users from around the world are developing accelerated applications. Here is a sampling of their publications:

Real-time Quantitative Phase Imaging of Red Blood Cells – Zhejiang University and Stanford University

Dissipative Dynamics at Conical Intersections – Nanyang Technological University, Singapore

Dynamical Properties of a Nonlinear Growth Equation – Université de Picardie Jules Verne, France and the University of Miskolc, Hungary


ArrayFire Community Meetings

The next community meeting is on October 17, 2022. An invitation to this meeting and a review of previous meetings are available on Github.

Job Openings

GPU Computing Software Engineer (CUDA, OpenCL, oneAPI) – Apply Here

  • Offices in Atlanta, Georgia. Also available remotely.
  • Full-time and part-time positions are available.

Consulting and Training Services

ArrayFire is free and open source and always will be.

ArrayFire serves many clients through consulting and coding services, algorithm development, porting code, and training courses for developers. Contact us at sales@arrayfire.com or schedule a free technical consultation to learn more about our consulting and coding services.


Newsletter Signup

Leave a Reply

Your email address will not be published.