Why isn't OpenGL ES 2.0 / WebGL sufficient for most games and game engines?

I've seen a number of impressive WebGL demos and games running on different browsers and operating systems as well as even mobile browsers (My 3rd generation iPad can run a number of older WebGL demos at 30+ fps).

Since WebGL is based on OpenGL ES 2 which is a subset of OpenGL 2 (OpenGL 2.x and OpenGL ES 2 both date back to the mid-2000's) why do many game engines still offer multiple rendering systems (DX9, DX10, DX11, OGL 2.x, OGL 3.x)?

 

Well, as far as most game engines offering different APIs, it is simply that some work better than others in different situations and needs. For example, on current hardware on windows, if you want tesselation shaders, your choice will be DX3d11, as there is no other current choice. I think the latest OGL can do it, but I'm not sure if it simply isn't ready yet or not supported by hardware yet. On the other hand, Nothing but windows(and XBOX and windows phone) support anything DX3d, so you have to use OpenGL or a software renderer there. Also, some hardware, though it supports a higher version, may well be faster with lower versions. Or simply the hardware that isn't the video card isn't advanced enough to utilize features of newer API versions even if the video card can.

So yes, it is mostly due to wanting to support both older hardware as a fallback, and have features of newer hardware, all in the same game engine.

 

But how much can the average game engine or game developer actually benefit from incorporating the features found in newer versions of DirectX and OpenGL?

I can only imagine a very small percentage of game developers actually needing features like 64-bit floating point, hardware tessellation, instanced arrays, or compute shaders. Not for the vast majority of games being made these days (i.e. infinite runners, clickers, pixel art games, zombie apocalypse shooters).

Also, Microsoft officially supports ANGLE (OpenGL ES 2.0 emulation layer for DirectX) which makes a separate DirectX game-engine renderer redundant.

https://github.com/MSOpenTech/angle

 

Games are often about performance. Using the most efficient API for your target system often plays a role in that. Sure, you could technically get the same pixels drawn regardless of which API you use, but how much longer will some tasks take on some systems?

 

Well, as an example, Unity has a target market that includes not only that average developer that doesn't need the new effects, but the crowd(yes, I said crowd, as in sizable group) that wants those new features for their games. Many devs, when given the chance, will indeed use those features, even if they don't actually need them, in order to make the games better if the game player has a system that can handle it. That is in fact the beauty of game engines like Unity(and Irrlicht, as it would do the fall backs too, though it isn't really modern or relevant anymore), that they can do the new features on newer hardware and pretty seamlessly fall back to not using them, downgrading materials, etc... on older systems as needed, and even allow you to emulate those systems' capabilities even on newer hardware. That last feature isn't perfect as it doesn't truly emulate the speed of those older systems, but at the least the rendering models, materials, shader levels, etc..., can be emulated by Unity.

 

But many games don't benefit, at least meaningfully, from the performance gains provided by more efficient API's. For example, the average iOS or Android game written in C++ can be ported to Windows Phone using Angle without any major performance drawback, something advantageous with Windows Phone's small market share.

Given the likelihood of bugs and lack of support for cutting edge graphics cards used by much fewer people, are these cutting edge features long-term beneficial?

 

You keep using the term "average" as if it means the same thing as "all".

 

I never said or implied "all". Even if that were the case, you could've presented anecdotal evidence in which one or a few games, even some flappy bird or 3d ping pong clone, actually did suffer when ported to platforms using seemingly less efficient APIs, and that would've been more beneficial to your position.

 

You can view the benefits from many different angles... some API's make it easier for developers to build certain systems, that saves the company time and money which later on makes it easier for them to brake-even.

Some API's give you access to exclusive functions, often hardware stuff... this might make your product a lot more interesting if a majority of your target group is sitting on the mentioned hardware.

Some languages, API's and platforms are faster or a combination of these might be better optimized and work better together. The amount of speed that you need depends on the project... a person that wants to make a multiplayer tic-tac-toe could make it run on ancient mobile devices etc... but there are lots of technologies that developer want to use but they can't because we don't have the power to run them in real time and new technologies will arrive by the time that we can those that we wished for in the past.
Also, this an intentional situation created on certain platforms... Microsoft want you to use DirectX and make you optimize the game for it because that makes the Windows platform the most attractive place for customers to enjoy the product.

Wut? Sorry but what?

Cutting edge technology will be tomorrows dust in the mirror. Intel is just trying to brake into the graphics market if that is what you are thinking on but a serious gamer would still get a computer with proper AMD or INVIDIA card... Why do you think we get new consoles every once in a while? You have to know your audience and product to know how far you should target tech. Also, using a cutting edge feature might be a wonderful marketing argument.

 

Cutting edge graphics used by much fewer people? What are you talking about? NVIDIA has stated they've seen an increase in earnings of 31%. More than they were even expecting.

It isn't surprising given the recent influx of 4K/5K monitors and needing hardware capable of driving them at acceptable speeds. Only the cutting edge cards can do it fast enough at the moment.

http://www.wsj.com/articles/nvidia-profit-rises-31-1423691605

 

Well, I heard the newest NVIDIA Tegra's (X1) will support OpenGL 4.5 and DirectX 12.

Given the near ubiquity of OpenGL ES 2.0/WebGL capable hardware, I think this is total waste of API and graphics capability for a platform dominated (largely, not completely) by 3d ping pong and flappy bird clones, not to mention the "retro" zombie shooters that look like they were made in 1993.

 

Are mobile devices used solely for those purposes though? Shouldn't newer hardware consume less power due to having to spend less time rendering and at a lower clock rate? Don't forget too that mobile devices are not as low resolution as they used to be. That 7-inch screen can often be higher resolution than some big TVs.

 

I didn't say you did. Your main reasoning for taking the approach you're asking about is that you don't think "most" games would suffer for it. That's quite possibly true. My point is solely that "most" isn't "all", and it's not a good idea to write things off just because "most" won't need them.

Also, whether people need them or not doesn't necessarily correlate with whether they want them and/or whether or not they'll impact engine choice.

What does "major" mean? It implies that there is a performance drawback. Also, what about those using Metal?

 

This is the best explanation, by far, thank you.