The cloud is the next big thing, and NVidia is looking to capture market share with a cloud offering specifically for gaming. It is a good strategy, because there is a big need for this type of cloud service. The video gaming industry is a $20 billion industry, and video is a key hardware component in gaming. On average you must have a GPU that retails for $300-$500 to get a decent gaming experience, and that needs to be refreshed about every 2-3 years to keep up with the gaming technology.
The average modern video card has a GPU with about 1,000 cores. To refresh memories, CPU’s have 4-8 cores. Yes, the GPU is hundreds of times faster than your Intel Core I-7 CPU. It takes that much compute power to render a 3D world at 30 frames-per-second, with each frame having hundreds of polygon mesh 3D objects on the screen with each object having hundreds of thousands of vertices. Your GPU has to compute all these vertices in 3D space, relative to the player’s viewpoint, and fill in the faces with textures, compute and apply lighting, shading, etc.
That is why a computer with a $75 video card can play streaming video without issues, and choke on a 3D game… because the GPU has to calculate each frame in realtime. This is where the GPU cloud comes in — because if the 3D rendering can be offloaded to a cloud service, then everyone can enjoy games and apps that demand $500 video cards with $75 video cards.
The typical 3D game tracks the players position/viewpoint, the millions of vertices that make up the polygons that define the world and currently in the scene. The orientation of all those vertices from the perspective of the players viewpoint are calculated by the GPU, and then the faces of the polygons are filled in with the textures. All this is done by your high end GPU…. but in the future would be done by the gaming cloud instead.
To achieve this, games would be written to be cloud friendly, and a module would run on the cloud which knows the game video make-up: the objects in the world, the polygons/vertices that make up each, and the textures that are needed for the polygon faces. All of this is stored on your PC today taking gigabytes upon gigabytes of space, because the graphic rendering is done by your PC. All of this will be offloaded to the cloud, and the local game engine would simply transmit your position, and the position of other objects up to the game cloud, and the rendered video comes back just like streamed digital content.
There is an additional layer of complexity in 3D rendering — light sources. I left that out in the above for simplicity. But the rendering of a scene is also impacted by the objects that emit light, such as sunlight in the scene, headlights on cars, etc. The gaming cloud would also know the lighting objects in the world/scene in order to also apply lighting/shadowing to the scene just as your local GPU does today.
This is all very doable, and in the works. NVidia is working on it, and Amazon has some GPU specific cloud offerings right now.
This is exciting and will have a huge impact on the gaming industry because we will see the graphics capabilities in games take huge leaps forward. The reason is simple — game developers have to develop the game for a hardware platform that the majority of the user base can afford in their homes at the current time. About every 5 years we see games that up the ante and offer the next generation of video complexity/detail in a title; but when that happens the studio is bombarded with complaints and criticism because a good segment of the user base doesn’t have the hardware required to run the game. When this happens, it is very costly and damaging to the studios industry reputation, so it makes sens that studios will try to avoid huge leaps forward in order to steer clear of that pain.
It is hard to predict what the business model will be with gaming in the cloud. I would expect the model to be some type of monthly charge based on use time and GPU compute power. Will that charge go to the end user, or will the studios eat that cost and roll it into the price of the title?
One barrier that exists to transitioning to cloud gaming is that people who do have high end GPUs today will not want to pay for a cloud service because they don’t need it, having already made the investment in high end GPU hardware. Because of this I would anticipate games will be released that offer local or cloud GPU support, which will allow the consumers that have the hardware to run it to get their ROI, whilst allowing those who do not hard the GPU hardware to run the game without a hardware investment.
Console systems (Playstation, XBox, wii, etc) will come down drastically in price, and possibly be obsoleted. The graphics hardware is the biggest driver of the cost of the system. Once graphics are offloaded to the cloud, the console system will become a thin client streaming device.
Gaming on mobile devices (Andoid/iOS) will be much improved. Mobile devices have serious constraints that create a barrier to good gaming.
- Tight power consumption constraints as a tablet that only lasts 2 hours on battery won’t sell, and that means advanced graphics compute power isn’t possible.
- Tight hardware cost constraints, because the price point for the average table is no longer $600, which also means advanced graphics capabilities isn’t possible.
- Tight storage constraints, because the average mobile device has 16GB of storage space, which means games cannot have 20GB of textures stored on the device.
Cloud gaming will alleviate all these constraints and you will be able to play the same titles you can play on a PC or console.
Lastly, the losses in revenue due to piracy will be alleviated and the headaches of digital rights management (DRM) will no longer be necessary. People will not be able to play games they did not purchase because most of what the game does is offloaded to a cloud. People who don’t pirate games won’t have to pay a slightly higher price in order to offset the losses suffered by those that do pirate games.
Happy Gaming!
Tom C