Slashdot Mirror
What Valve's Announcements Mean for Gaming
Now that we have the full picture of Valve's efforts to bring PC gaming to the living room (SteamOS, dedicated hardware, and a fresh controller design), people are starting to analyze what those efforts will mean for gaming, and what Valve must do to be successful. Eurogamer's Oli Welsh points out that even if Steam Machines aren't able to take the market away from Sony, Microsoft, and Nintendo, they put us a step closer to the final console generation. "Valve has hopefully sidestepped the most depressing aspect of console gaming: the enforced obsolescence that makes you consign your entire games collection to a dusty cupboard every five years." GamesRadar notes that Valve's approach is fundamentally different from that of the current console manufacturers because it's about putting more power into the hands of the users. "The takeaway from SteamOS, then, is that openness breeds innovation. Valve's putting the very source code of its operating system in the hands of everyone who wants it just to see what happens. Comparatively, Microsoft is pushing its Windows Store, turning Windows into an increasingly closed platform (i.e. one that charges costly development licensing fees and restricts access to certain content providers)." Everyone's curious to see how the controller will perform, so Gamasutra and Kotaku reached out to a number of game developers who have experimented with prototypes already. "[Dan Tabar of indie studio Data Realms] said the configuration map for the controller allows you to do 'pretty much anything.' For example, developers can slice up a pad into quarters, each one representing a different input, or even into eight radial sections, again, each section representing whatever you want, mapping to key combinations, or to the mouse." Tommy Refenes, co-creator of Super Meat Boy, wrote an in-depth description of his experience with the device. He summed up his reaction by saying, "Great Start, needs some improvements, but I could play any game I wanted with it just fine."
SteamOS has one MASSIVE problem- performance in AAA gaming. SteamOS (at this time) can only offer the horribly broken/inappropriate OpenGL and OpenGL ES family of drivers/APIs. The OpenGL family, as well as being only somewhat similar to DirectX, varies in performance to extremes when considering revisions of GPU hardware from AMD, Nvidia, and Intel. It is a dirty little secret that these three hardware companies only want to create 'good' OpenGL drivers for their latest GPU designs, often disowning parts that are only a few years old (Intel, for instance, is NEVER going to have good drivers for the Atom chips that use PowerVR GPU cores).
Now, when it comes to 'casual' games, most new GPU parts have so much power, they are overkill for such titles, and can deliver good gaming performance even with appalling inefficiencies in the drivers. Hoverer, even casual games are starting to leverage the excellent deals AAA engine companies offer developers who use their products in modest (saleswise) applications.
The bottom line is that SteamOS needs to have next-gen console ambitions. It needs (in the better Steam boxes) to desire to compete with the Xbone and PS4. There is only ONE way this can happen- by complete adoption of AMD's Mantle driver/API on SteamOS. Unlike DirectX, Mantle is OS agnostic. Mantle (or something very like it) will be the low level API for all development on the Xbone and PS4 (both of which have 2014+ PC-like architectures).
Mantle requires the computer (be it console or 'desktop') to use x86 CPU cores, and AMD's GCN GPU cores. GCN can be thought of as an ISA (instruction set architecture) like x86 or PowerPC or ARM or MIPS. Take the x86. The internals of the CPU changed massively from 386 -> 486 -> Pentium -> Pentium Pro/Pentium 2 -> Pentium 3 -> Pentium 4, BUT the 32-bit x86 ISA remained the same. AMD will likewise update the internals of their GPUs generation on generation, but the GCN ISA will remain.
Mantle allows developers to talk to the computing units of the GPU directly, something OpenGL and DirectX can NEVER, EVER do. Therefore, in future situations, Mantle may be 10x+ more efficient at KEY rendering tasts than OpenGL/DirectX- an advantage no next-gen AAA console developer would dream of ignoring. Most AAA games will originate for the console market, and will NEED Mantle to allow ports to PC like platforms, including Steam Boxes.
Mantle solves the issues of lousy/untrustworthy/problematic graphics drivers on Linux in one simply, incredibly price efficient act. The fact that it locks one vendor in (AMD) and one important vendor out (Nvidia) is an issue, but Nvidia is essentially copying AMD's fundamental approach to the GPU in their 2014 GPU designs, and could possibly come to an arrangement to make their parts GCN 'compatible' (like how AMD makes x86 CPUs compatible with those made by Intel).
Anyway, if SteamOS adopts Mantle for AAA, and OpenGL ES2.0 and OpenGL ES3.0 for more casual games, there is no reason Valve can't have a fantastic success with this project. Following the Microsoft path on the other hand (forcing extraordinary hardware overkill to merely maintain 'parity' with the next-gen consoles) will doom SteamOS.
Nice, due to this we'll have Samsung and Huawei games consoles. Give it all away.
Indeed, the AMD API they are proposing is supposed to be much lower level than OpenGL because console developers are used to getting away with being much closer to the metal when they only have 1 hardware target to shoot at. To get an idea of this, in neither OpenGL nor DirectX is there an efficient method of just taking a pointer into video memory and fucking around from the CPU side of things.
But when you look at AMD's APU setup, memory is memory.. video memory and main memory are one and the same.. there is no reason that you shouldn't be able to just go ahead and write to individual texels in a texture efficiently, and so forth.. something quite inefficient on a PC with a dedicated video card in the x16 slot.
AMD plans for this API to be "open" so Intel will be free to implement it on their integrated GPU's as well... NVidia, without its own x86/x64 architecture, will be screwed of course.
"His name was James Damore."
Except that mobile gaming is a completely different beast from living room or PC gaming. Mobile gaming is about quick bites, simple controls, and shallow gameplay (this isn't a bad thing, per se). Mobile gaming is casual, by default. Its hard to get into an epic RPG while on the bus, or in the dentist's office. If I'm going to play something like Skyrim, I'm going to do it in a comfy chair, on a good screen, with mature controls.
Tethering a controller to your phone or tab is counterproductive, since you "un-mobiled" mobile gaming, by forcing someone to carry around a controller as well as their device.
Mobile isn't replacing anything, I wish that fallacy would die. Mobile is supplementing a certain part of traditional markets, but it isn't replacing the core of those markets. Looking at console and traditional game sales back this up, they aren't slowing down in relation to rise in mobile device sales. Nor will they, since they fill a very different niche than traditional consoles and PCs for gaming.
Same with the stupid trope that mobile will magically kill traditional PCs... This is said by people who never used their PC for anything more serious than email and light web browsing. There is very little in my daily computer tasks that can be moved to mobile, outside of light email and web duties. Sure, this is a gap MS is targeting (badly) with the Surface Pro, but suddenly we're not talking mobile anymore, but a traditional laptop with a floppy keyboard and optional touch controls. And still it isn't going to be as good as my large screen for most tasks.
The living room died so many decades ago
I'm now picturing a family of four huddled in their backyard streaming watching movies on a 10" tablet. I feel kind of bad for them, since they could be inside, sitting in their living room watching it on an increasingly affordable giant HDTV.
A patriot must always be ready to defend his country against his government. -edward abbey
Valve wants native ports, not Wine usage. Their strategy for streaming is sort of a Microsoft tactic. Consumers will find Steaming to be cumbersome and eventually demand native ports from publishers, that's valves goal with streaming; Native ports. They've had plenty of time to figure all this out and gather the data. It's a big risk for them, they aren't just going to throw something out there and hope it catches.
In the end we all need to stop thinking about Wine. It's more of a problem, than a solution.