At CES when we unveiled the first details of our new GPU Vega architecture –codenamed “Vega” — I was hoping for a positive reaction, but I didn’t quite expect this. Millions of gamers, game developers, data scientists, and professional content creators around the world are anxiously awaiting products based on Vega. And every week, I get hundreds of emails and tweets from community members asking me when these products will be available on store shelves.
Well, it’s time.
Today we’re announcing the first of several Vega architecture-based products to roll out this summer, and soon we’ll be launching Radeon Instinct, Radeon Pro, and Radeon RX flavors of Vega. But today’s Vega is just a little different. Let me tell you why.
Radeon™ Vega Frontier Edition
|Single precision compute performance (FP32)||~13 TFLOPS|
|Half precision compute performance (FP16)||~25 TFLOPS|
|Pixel Fillrate||~90 Gpixels/sec|
|Memory capacity||16 GBs of High Bandwidth Cache|
|Memory bandwidth||~480 GBs/sec|
What comes to mind when you hear the word “pioneer?”
At AMD, we think about the people who risk everything in the name of progress. These modern-day pioneers embrace and propel new technologies forward. They work to heal our own planet while fearlessly exploring new ones. They harness science to fuel creativity, and employ creativity to drive science. Pioneers look at boundaries and see only a starting line. Their achievements are measured not in years, or even decades, but in centuries.
We’ve spent years preparing to enable this next generation of data scientists, immersion engineers, and product designers, by designing a formidable new GPU architecture – Vega – from the ground up, and by heavily investing in machine learning, high performance graphics for content creators, and open software solutions to deliver value to a broad pioneer ecosystem. Today our vision for delivering the future of data science and advanced visualization materializes with the release of the Radeon™ Vega Frontier Edition graphics cards which will be shipping in late June.
We’re excited about releasing this graphics card into the wild and seeing what you’ll accomplish with it, especially with the horsepower that’s inside, because for machine learning and professional visualization workloads, Radeon Vega Frontier Edition is very simply the fastest graphics card on the planet.
At our product’s core is Vega’s disruptive High Bandwidth Cache Controller, the cornerstone of the world’s most advanced GPU memory architecture. Combined with HBM2, Radeon Vega Frontier Edition helps expands the capacity of traditional GPU memory to 256TB, allowing users to tackle massive datasets with ease. The result is a graphics card that can easily handle demanding design, rendering, and machine intelligence workloads.
Radeon Vega Frontier Edition Vs. Fiji Architecture
But hardware is only half of the story – a significant part of our investment and resource commitment is dedicated to fostering an open software ecosystem. We continue to aggressively improve our ROCm open software platform and machine learning libraries. We’re also supporting open machine intelligence frameworks like Caffe (released in April). Later this quarter we plan to offer support for Torch, and Tensor Flow is in the works.
For Data Scientists, Immersion Engineers, and Product Designers
The advantage of the Radeon Vega Pioneer Edition graphics card is unmatched performance in a remarkable range of use cases. Here are a few examples:
The Radeon™ Vega Frontier Edition, combined with our ROCm open software platform, delivers industry-leading performance in deep learning. Developers can now tap into the power of the Vega architecture to do machine learning algorithm development before deploying it to massive servers that will be equipped with Radeon Instinct accelerators. In the initial testing (early clocks and drivers) we did against the fastest competitive card we could get our hands on, we found Radeon Vega Frontier Edition graphics cardsto be more than 30% faster as measured by the DeepBench Benchmark 1 – and also delivers a disruptive performance per dollar equation that enables AMD’s leadership in compute TCO for the foreseeable future.
Advanced technologies like real-time visualization and physically-based rendering give design and manufacturing firms a powerful arsenal of visualization tools that can be used at all stages of product development. The Radeon Vega Frontier Edition graphics card provides the necessary performance required to help drive these increasingly large and complex models through the design phase as well as rendering phase of product development.
SpecViewperf 12.1 (higher is better)
Revolutionizing Game Design and 360 VR Workflows
The Radeon Vega Frontier Edition graphics card simplifies and accelerates game creation by providing a single GPU that is optimized for every stage of a game developer’s workflow — from asset production, to playtesting, to performance optimization. With AMD graphics technology found in many of today’s game consoles and PCs, developers can build and optimize next-gen games for a large install base. And combined with Radeon Loom, AMD’s revolutionary 360-degree video stitching technology, a new breed of filmmakers will be able to harness the compute power of Radeon Vega Frontier Edition to stitch high-resolution video in real time.
The Radeon Vega Frontier Edition graphics card is going to empower the pioneers creating the next generation of gaming experiences, but it does beg one question: Can you game on a Radeon Vega Frontier Edition? The answer is yes, absolutely. But because this graphics card is optimized for professional use cases (and priced accordingly), if gaming is your primary reason for buying a GPU, I’d suggest waiting just a little while longer for the lower-priced, gaming-optimized Radeon RX Vega graphics card. You’ll be glad you did.
On behalf of everyone at AMD and Radeon Technologies Group, thank you to the pioneers of today and the visionaries of tomorrow. To learn more about the Radeon Vega Frontier Edition graphics card, visit http://radeon.com/frontier and we’ll be providing further details by mid-June. I’ll also be hosting an AMA on Reddit this Thursday at 2 pm PST – please join me at reddit.com/r/AMD.