Slashdot Mirror
SteamOS Gaming Performance Lags Well Behind Windows (arstechnica.com)
New submitter NotDrWho writes: As reported by Ars Technica: "With this week's official launch of Valve's Linux-based Steam Machine line (for non-pre-orders), we decided to see if the new OS could stand up to the established Windows standard when running games on the same hardware. Unfortunately for open source gaming supporters, it looks like SteamOS gaming comes with a significant performance hit on a number of benchmarks." They tested with two graphically intensive titles from 2014, Middle-Earth: Shadow of Mordor and Metro: Last Light Redux. They say, "we got anywhere from 21- to 58-percent fewer frames per second, depending on the graphical settings. On our hardware running Shadow of Mordor at Ultra settings and HD resolution, the OS change alone was the difference between a playable 34.5 fps average on Windows and a stuttering 14.6 fps mess on SteamOS." Even most of Valve's own games took big performance hits when running under SteamOS.
On one machine, in two games.
I recognize that testing this sort of stuff on a wide variety of hardware and with many games is hard, and that they haven't had the time yet to put together a thorough analysis. But you should really qualify your results, like "preliminary testing has indicated that Steam OS performance may be worse than Windows 10 performance in some games on certain hardware configurations."
But that makes for a terrible headline :p
One of my primary suspects for the difference is the video card - how well optimized are the Linux drivers?
Not true - in fact, Nvidia's Linux driver is quite good. The issue is that 'important' games get special attention from the graphics companies, who special-case things in their drivers - replacing whole shaders, etc. That doesn't happen in Linux. It winds up being necessary because OpenGL has grown so complex that it's incredibly hard to write fast code for it.
Vuikan is liable to change that considerably - a much lower-level API, that engines can interface with more directly and consistently. The drivers won't have be huge tangles of special-case code, and will be much simpler to implement on multiple operating systems because they are called upon to do far less.
PHEM - party like it's 1997-2003!
Like I said, I'd love to be wrong, and I'd LOVE my gaming PC to be running SteamOS instead of Windows. But I want to play Mass Effect and Fallout and Skyrim and Tomb Raider and Battlefronts and Uncharted and Assassin's Creed.
That's MY definition of a Casual Gamer - I get a relatively low number of games, but I play the hell out of them. I switched to PC because consoles became a really unfriendly environment for people like me, and Steam is just way the heck off. To play the games I want on PC right now I need both Steam and Origin. And if EA /ever/ lets their stuff go on Steam, well. It ain't going to be any time soon. SteamOS might be an OK platform eventually, but it's far from it now, and won't get to "OK" for a bit, either.
And, again, I would love for it to succeed, but they've been at this for years.
In the land of the blind, the one-eyed man is kinky.
"If that's true, then why aren't game developers focusing on OpenGL instead of DirectX then?"
Because they already know direct x and probably don't know opengl. Because direct x is what is primarily targeted and optimized by graphics card drivers with opengl a secondary consideration. And lastly because sound is then organized into the same library whereas with opengl they would then need to pick a secondary sound library.
OpenGL is an api, it has no performance characteristics, those all come down to the implementation.
"Are you saying that every developer in the industry is in a mass conspiracy with Microsoft"
No conspiracy needed. Microsoft has no interest in opengl performing well on windows therefore they don't expend any effort to make it perform well and thereby cripple it relative to their own library. Manufacturers target Direct X because game developers do, game developers target Direct X because manufacturers do.
In almost every case the "ridiculous big conspiracy" argument is a false dichotomy, you don't need active conspiracy for individuals industry wide to create the same result active conspiracy and collusion would. In most places this occurs out of lots of individuals working out of self interest. The same is true of widescale consumer screwing in industry, if all the major competitors engage in the same practice the consumer can't vote with their dollar, the major competitors don't actually need to agree though. The same things are profitable for all of them, if they all pursue what is most profitable for themselves the result will be all of them engaging in the same practice and screwing the consumer.