Source Engine Faster On Linux

8676d29610e6c98d6dd2d9c38528cd9c
0
alphadog 101 Aug 04, 2012 at 14:14

As Microsoft begins to wall-garden itself, Apple-style, with Win8, this seems to be another step down that road:

http://www.extremetech.com/gaming/133824-valve-opengl-is-faster-than-directx-even-on-windows

Linux/OpenGL/L4D2: 315 fps
Windows 7/OpenGL/L4D2: 303.4 fps
Windows 7/DirectX/L4D2: 270.6 fps

Discuss… :)

7 Replies

Please log in or register to post a reply.

6837d514b487de395be51432d9cdd078
0
TheNut 179 Aug 04, 2012 at 15:02

I read Gabe’s opinion about Win8 and Linux, but I didn’t think he was in the works to port the source engine. Good on them. As for the benchmarks, meh ;p I think the important thing is that Valve is adopting OpenGL and other platforms. Whether or not people will adopt it solely on the presence of more games is debatable. I love developing for Linux and using it for some things, but it has a lot of hurdles to overcome if it is to become a viable alternative to the average Joe. It’s a question of whether to improve Linux or improve the technical literacy of the general public :) Novell tried to pull that stunt back in 2K and it backfired.

Still, this will be interesting to see where it goes. I’m not crazy for Win 8 since I’m quite content with 7, but it will be interesting in the next couple of years where things go. With C++ and OpenGL, you open up PC Windows, Linux, Mac OS, iOS, and Android. With C++ DX you have Windows, XBox, and Win Phone, the later which hasn’t shown a lot of promise yet (sadly). If Android and iOS secure their future as market leaders, pushing Windows mobile out, then that too will be a factor, I think, when people decide which APIs to use. The game is afoot. Let the GL vs DX wars commence :)

A638aa42130293f319eda7fa4ba121f4
0
fireside 141 Aug 05, 2012 at 20:05

I’m thinking about getting a Raspberry Pi. Looks pretty cool. I might go back to 2d with html5. Linux will probably always be a niche for the desktop. People will grumble but stick with Windows. Many will probably skip Windows 8 like they did Vista. It will make it harder for OEM’s to make a profit.
Valve’s engine should be a shot in the arm for Linux anyway.

820ce9018b365a6aeba6e23847f17eda
0
geon 101 Aug 05, 2012 at 20:38

@fireside

shot in the arm

Like a shotgun or an injection?

A638aa42130293f319eda7fa4ba121f4
0
fireside 141 Aug 06, 2012 at 00:08

Maybe that phrase isn’t used much anymore. It’s an injection:

http://idioms.thefreedictionary.com/shot+in+the+arm

99f6aeec9715bb034bba93ba2a7eb360
0
Nick 102 Aug 07, 2012 at 00:05

@alphadog

http://www.extremete…even-on-windows Linux/OpenGL/L4D2: 315 fps
Windows 7/OpenGL/L4D2: 303.4 fps
Windows 7/DirectX/L4D2: 270.6 fps

Seems kinda pointless and insignificant to me. I mean, as mentioned in the article, their initial version ran at just 6 FPS, and so obviously it required lots of optimization. But they could easily overshoot the Windows results that way, by chance (or intentionally). This old engine was never optimized to run at 300 FPS on Windows a killer system like that. It was optimized to run at 30 FPS on a GMA 950 or something like that. I’m fairly confident that optimizing it on Windows, again, would result in similar or higher framerates (if they really wanted to).

So I’d be a little skeptical about attributing these different results solely to the operating systems. I certainly don’t want to exclude the possibility that Linux is indeed more efficient, but the above data is anecdotal at best and not good science.

A8433b04cb41dd57113740b779f61acb
0
Reedbeta 167 Aug 07, 2012 at 00:42

Also, keep in mind how high all of these framerates are…we’re talking about a 0.5 ms difference between the slowest and fastest cases, or about 3% of a 60Hz frame. It’s nothing to sneer at, but it’s a pretty small difference in real terms.

(Hey, Nick! Good to see you again. :))

8676d29610e6c98d6dd2d9c38528cd9c
0
alphadog 101 Aug 07, 2012 at 19:25

@Nick

So I’d be a little skeptical about attributing these different results solely to the operating systems.

Agreed. Another general issue that has always bothered me about many systems websites that review hardware is the lack of “error bars” in their benchmarks, a.k.a. standard deviation. You don’t know if there’s a 3-5fps swing in their data, or a 30-50fps swing.

OTOH, I don’t think you have to totally dismiss it because “it’s not good science”. It has its own value. It does demonstrate something possible, rather than falsehoods. For example, do you think L4D2 is not optimized for Win7/DirectX and that those score are not representative of that platform?