What Radeon Instinct Means For Gamers

  • Share:
  • Facebook
  • Twitter

Machine Learning is the next big shift in technology that will profoundly change the way we live our lives.

I figured it would be awesome to start my first ever Radeon.com blog with an attention-grabbing statement–I hope I have your attention now!

I should follow such an optimistic statement with a bit of a history lesson. Artificial intelligence has been a mainstay in popular science-fiction media for years. If you bring up AI in popular media—whether it’s that kid seeing dead people acting as robotic replacements, or evil robots farming humankind for power in the movies – there is a doomsday or serious ethical consideration not too far off.

This tendency for our imaginations to focus on such worrying scenarios could be due to the common understanding that AI, whenever it does become ubiquitous in our daily lives, will have profound effects on the way we live. But profound doesn’t have to imply scary!

The optimism surrounding machine intelligence has been building steadily over past couple of years, and some of that excitement has likely reached you through the news and pop culture. Why the sudden uptick in discussions about this technology?

In short, some theoretical methods that computer scientists have researched and written papers about for decades have finally become practical through the use of GPUs.  The AMD ROCm platform, for instance, unlocks the potential of Radeon Instinct accelerators much like DirectX®, OpenGL®, and Vulkan™ have done for gaming and game developers.

As a gamer, you probably have two nagging questions on your mind as far as AI is concerned:

  • How soon can you upgrade your rig to include a Radeon Instinct accelerator?
  • How will machine intelligence technologies affect gaming?

Should you include a Radeon Instinct accelerator in your next gaming build?

I don’t blame you for wanting a Radeon Instinct accelerator in your gaming rig. The boards are beautifully designed and would certainly look at home alongside the rest of my system’s components!

Unfortunately, I have to be a bit of a buzzkill: if you are a gamer with no interest in researching, implementing, or designing machine intelligence algorithms, then Radeon Instinct accelerators are not for you. These GPUs are passively cooled and they do not have any display outputs—a couple of major complications, as far as practicality for your gaming rig is concerned!

How Radeon Instinct accelerators could affect the future of gaming:

The truth is that the game development industry has been at the forefront of the practical application of AI concepts for years.  Here are few examples:

  • Your RTS unit that got caught on some terrain and never made it to its assigned destination? A crude form of path-finding that borrows concepts from artificial intelligence research.
  • That somewhat-predictable spell tactic used by a computer-based player in your favorite RPG? Likely part of a very crude decision tree of logic implemented by game developers.  Still, this logic could be considered “artificial intelligence”.
  • What about the easily embarrassed computer-controlled goalie in my late-90’s ice-hockey games? Calling that AI is likely a stretch, but you get the point!

We are only at the beginnings of integrating AI with our day-to-day lives, but the possibilities for gaming are truly exciting. By no means is this an exhaustive list, but these are just some of the ways gaming will benefit from the utilization of machine intelligence:

  • Realistic worlds: Imagine a world that is fully interactive. Every NPC has unique things to tell you and is all-knowing about your decisions and habits in the game. Or how about an RTS where you have to train your units’ AI systems and you end up putting their capabilities to test against your friend’s units?
  • Smarter opponents: No more predictable spell patterns, no more easy goals in ice hockey games, and say goodbye to a lack of replay-ability! If each enemy in a game fully utilized a Radeon Instinct accelerator in the cloud to make intelligent decisions about how to beat us in our games, the possibilities for ragequits will be endless! The truth is that a lot of early practical machine intelligence work was conducted with gaming.  Whether it was solving chess, checkers, or Go, it’s clear that when you can’t beat them, the next best thing is to build a crazy expensive computer that can!
  • Input methods: Consider the difference between a traditional keyboard-and-mouse gaming setup and the input used by Microsoft’s Kinect®.

Microsoft Kinect was made possible by training a machine to conduct real-time image recognition and identify a wide variety of bodies and their movements. The more machine intelligence becomes a part of our day-to-day lives, the higher the likelihood that we expect to interact with computers, and by extension our games, in a more intuitive fashion. I love the keyboard-and-mouse as much as the next gamer, but I can imagine a future where our love affair with this setup is surpassed by a weird mix of eye tracking, brain scanning, and voice recognition techniques based on machine intelligence.

All of these ideas are just off the top of my head, sparked by discussions with some of Radeon Technologies Group’s brightest minds, progressive thinkers and forward-thinking builders.

Be sure to visit instinct.radeon.com for more and leave a comment below if I’ve missed a glaring opportunity for gaming and machine intelligence. I’m already struggling to keep up with the current complexities of game design, and with machine intelligence in the pipeline, my years of being a capable gamer are numbered.

Even so, as far as gaming is concerned: I, for one, welcome our new machine overlords.


Jason Megit is Technical Marketing Manager for the Radeon Technologies Group at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies, or opinions. Links to third party sites and references to third party trademarks are provided for convenience and illustrative purposes only. Unless explicitly stated, AMD is not responsible for the contents of such links, and no third party endorsement of AMD or any of its products is implied.

One thought on “What Radeon Instinct Means For Gamers

  1. I honestly think that if I were developing a game I’d be most interested in the Instinct cards for two use case. Because I’m a big Mechwarrior fan, I’ll present both use cases in the context of MWO.

    1) Hit detection. At present, there are certain mechs from the tabletop Battletech that can’t be implemented because the speed of those mechs simply breaks the hit detection of the Crytek engine as implemented. An MI25 card would be beefy enough to keep up with the ludicrous speeds of some of those mechs and register hits properly.

    2) Real simulation on a massive scale. One of the things about MWO is that all the maps behave as if they’re Earth standard. Autocannon rounds follow the same trajectory, lasers have the same range, etc; even on airless moons. I can imagine a 4P 32-core Zen server with 2xMI25 and 2xMI8 cards. The job of the first MI25 card would be to do continuous real time environmental simulations for maximum physics accuracy. The job of the second card would be to track and update in real time player based states (hit tracking and other things) for better and more in depth damage models. The two MI8 cards would be there to operate behind a blind client so that the AI would behave as a human would based upon what’s happening around them and respond accordingly. They would be trained by real Army vets who operated tanks and IFVs for maximum tactical realism.

    Btw, I’d love to have a couple MI25s for the yearly Overclock.net Folding War.

Leave a Reply

Your email address will not be published. Required fields are marked *