Date:
Monday , July 10, 2006
Author:
Brent Justice
Editor:
Kyle Bennett
Google +1

Effects Physics & Gameplay Physics Explored

Confused about effects physics vs. gameplay physics? We have asked ATI, NVIDIA, Havok, and Ageia about the two. We received in-depth responses providing great detail regarding effects and gameplay physics, what they mean for gameplay, and current hardware physics abilities.

continued...

ATI

ATI’s Godfrey Cheng, Director of Marketing for Platform Technologies responds.

(HardOCP) - Can you explain the difference between “effects” physics and “gameplay” physics?

(Godfrey Cheng) - Today there’s a lot of talk about gameplay physics vs. effects physics, and what one means over the other. What people are glossing over though is the significance of physics to gaming. Physics acceleration is the next big leap in gaming; it’s the next step in evolving the immersive game experience beyond what gamers know today. Just as highly detailed textures were a step forward in creating that immersive experience, just as post processing effects like motion blur and HDR were a step forward, physics acceleration is the next leap, regardless of whether the physics are “effects” or “gameplay”. That’s the central idea behind physics.

Aside from that though, as a general definition, “effects” physics include any element of the game that doesn’t have a direct impact on gameplay, such as waves in water, flowing hair or cloth, flying debris, and potentially colliding objects depending on how those objects are used. While effects physics may not have a ‘make or break’ impact on the gameplay, they do make games more visually interesting and realistic, creating a more convincing experience for gamers as things look closer to the way they would in real life.

“Gameplay” physics can also improve the look and feel of a game, but these physics may have a direct impact on the way a game plays. For example, your character’s weight within a given environment, or whether he simply falls through the floor or actually walks on it are both examples of basic gameplay physics.

The actual calculation of both types of physics is identical – both can be processed on ATI’s GPUs today. In fact, gameplay physics can be just as elaborate and detailed as effects physics, but what’s holding back developers from enabling gameplay physics acceleration on the GPU is that they want to ensure their games can be played by everyone.

Imagine a case where you blow up a wall in a game and thousands of pieces of debris fall to create a mountain that your character climbs to escape a certain area. In such a situation, every player of that game would need to have a system that supports GPU physics acceleration. If they don’t the game comes to a grinding halt because there’s too much information to process. For developers, this is unacceptable. The only fair way is to include elaborate effects physics only, and leave basic gameplay physics to be processed on the CPU so that the widest number of people can enjoy the game. That way, if you don’t have a system that supports GPU physics acceleration, you can still enjoy the game, you just won’t get quite the level of immersion you would with the effects physics turned on.

Just like other in-game settings, until physics acceleration has widespread adoption, game developers will provide physics effects as an option. For gamers who have ATI CrossFire graphics, that also gives them the choice of dedicating a GPU to physics processing, or harnessing both for rendering, so regardless of whether a game supports physics or not, you’re always making use of the hardware.

(HardOCP) - Are there any real-world gaming benefits to having gameplay physics accelerated with dedicated hardware?

(Godfrey Cheng) - Down the road when systems with GPU-accelerated physics are more common, there is a fantastic opportunity to transition more detailed gameplay physics onto the GPU. Not only will this be a benefit to gamers who will be able to play incredibly immersive games, but developers will benefit as they’ll still have a large addressable audience that they can market games to. For game developers, it doesn’t mean just an incremental improvement in game performance; instead it gives them the ability to include new game scenarios that have never been seen before. Imagine a scene where your character has to wade through a chest-high river that’s producing waves and a current, and your character’s movement reacts according to whether there’s a surge in the river for example. This type of realism will be possible.

(HardOCP) - GPUs allow effect physics right now, but technically speaking would it be possible to also use them to accelerate gameplay physics now, or in the future?

(Godfrey Cheng) - Absolutely. As I described above, the actual calculation of both types of physics is identical and from a technical point of view, both can be processed on ATI’s GPUs today. In fact, gameplay physics can be just as elaborate and detailed as effects physics. It’s simply a matter of ensuring that as large an audience as possible can enjoy a game. If only a small segment of the market has GPU-accelerated physics capabilities in their systems today, then developers don’t want to limit themselves to making games that only that audience can play. But as GPU-accelerated physics support becomes more mainstream, accelerating gameplay physics may be a more viable option.

(HardOCP) - Do effects physics put burden on the GPU lowering 3D graphics performance having to now render more objects on the screen?

(Godfrey Cheng) - Today’s GPUs are built to process hundreds of thousands of polygons, shading millions of pixels each frame. From a technical point of view, by adding more objects on screen you’re adding more polygons that need to be processed and rendered, so this obviously affects 3D graphics performance, but it’s no different than rendering the same number of objects without physics calculations.

Interestingly, games today are rarely limited by geometric complexity as a result of maximizing the vertex shader of the GPU. Because of the overhead associated with game objects in current APIs, often times games end up being CPU-limited. This means that the more objects you have in a given scene, the more the CPU is taxed, and you eventually start trading off frame rate as a result. So the vertex shader is hardly ever maximized because developers run into that CPU boundary first. So to sum up, that’s why developers traditionally have to consider the following: add more objects to the scene, make it more detailed and complex at the price of lower frame rates; or have less objects in the scene making it less complex at higher frame rates. Often the answer lies somewhere in between.

The same rule applies to effects physics, only on a different scale. As developers implement physics they’ll judge what is the most immersive physics simulation they can produce that still yields playable frame rates. Because of the massively parallel architecture of ATI’s GPUs, adding even 5,000 objects to a scene doesn’t result in a drop in frame rates. Bumping things up to 10,000 objects shows a slight drop in frame rates, but it’s nothing significant – the performance decline is not as linear as you’d see from the CPU. Today our early physics demos show that at high resolutions, we can simulate the collisions of 20,000 boulders running at frame rates well over 100 FPS.

As an interesting aside when talking about taxing the GPU with physics processing, physics acceleration actually allows gamers to get the full benefit of the X1K architecture. With physics processing, you have that geometric complexity that can push the vertex shaders to their limits, and the complex parallel processing needs that maximizes the dedicated branching processor of ATI’s GPUs. In essence, you put the whole architecture through its paces, which as a gamer; you like to hear as it means you’re getting the most value from your graphics card purchase.