Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Why is our graphics engine so old?
- Thread starter smali
- Start date
- Sort by reaction score
No I don't play that many games, there are only about three or four that I play regularly, but my point was that there's no noticeable difference between running them in OS X or Windows, despite the games being run with OpenGL. The person I was responding to made it sound like it was some horrible painful experience to play a game under OS X.
These tests were all wrong. I ran them myself on my MBP with the same GPU and roughly got the same performance as they did under Win7. The sole exception was Urban Terror, which remained slower. But since it's based on the same engine as open Arena (which ran great), I suspect bugs in this particular game.
Notice how they left the Mac mini with 1GB RAM.
What they measured is RAM swapping (which might be worse on OS X), not driver optimization. They didn't even notice they sometime got better frame rates at higher display resolution.
There is also a huge difference in how Windows handles graphics processing and how Mac does. In Windows the OS gives full control over to the application to give it the most processing power available. Sure there are still the OS calls and the DirectX/OpenGL calls but in general, the application now has the ability, to crash the system, lock it up, etc. BUT this is actually true when the game is running full screen, it is a totally different thing when running in a window; in this case the OS forces everything through it's normal processing channel.
In OS-X the graphics system runs under the OS regardless if it is full screen or windowed mode, everything is treated as if it is a window. This helps prevent problems with the application locking up or crashing and taking down the OS. In full screen, OSX just stretches out a window to maximize the space, but it is still running in a window. Unlike Windows that gives over more control.
A much better measure would be to run the windows game in a WINDOW vs. on OSX and test their speed. If you do this, you'll see a much closer performance, although I think Windows still wins out slightly. I did this with WoW at one time, and found about a 5 to 10% increase for windows and this was before the graphics system update. To do this correctly, you really need to have bootcamp installed and run the tests on one machine with two different operating systems. The games should have the same resolution set, both running in windowed mode and with the same settings.
In OS-X the graphics system runs under the OS regardless if it is full screen or windowed mode, everything is treated as if it is a window. This helps prevent problems with the application locking up or crashing and taking down the OS. In full screen, OSX just stretches out a window to maximize the space, but it is still running in a window. Unlike Windows that gives over more control.
A much better measure would be to run the windows game in a WINDOW vs. on OSX and test their speed. If you do this, you'll see a much closer performance, although I think Windows still wins out slightly. I did this with WoW at one time, and found about a 5 to 10% increase for windows and this was before the graphics system update. To do this correctly, you really need to have bootcamp installed and run the tests on one machine with two different operating systems. The games should have the same resolution set, both running in windowed mode and with the same settings.
Seriously, the performance difference between OS X and Windows remains basically the same in windowed and fullscreen mode. Plus, I don't think fullscreen mode in OS X is just like normal windowed mode. I remember RBarris (Valve developer) commenting on this at the Steam forums. Windows in Snow Leopard are not all equal, I think some have quicker access to the display, and this can change dynamically.
Most Linux 3D drivers (As of the latest X.org update) don't even have full 3D accelerated OpenGL 2.1 support (Including Proprietary Drivers). There is no comparison. Fun fact, nVidia override main components of the X window server to get full 3D accelerated support.
Mac OSX actually has full OpenGL3 support. What it doesn't have support for is GLSL1.3 (Which makes the OpenGL 3 support useless)
Also I question those benchmarks, not only as a Linux Advocate, a gamer and a games designer. It would've been a lot better to use a piece of software with good cross-platfrom optimization, like Doom 3.
OpenGL is "open source". Pretty much a collaborative effort for graphics that really belongs to no one....or perhaps a lot of "members".
DirectX belongs to Microsoft.
The first one...like Linux...is hard to make money on.
The second one...like Windows...belongs to someone who can be the sole beneficiary of marketing it. And incidentally, Micro$oft can pour lots of $$$ into it.
DirectX belongs to Microsoft.
The first one...like Linux...is hard to make money on.
The second one...like Windows...belongs to someone who can be the sole beneficiary of marketing it. And incidentally, Micro$oft can pour lots of $$$ into it.
I think what he means is that OpenGL was not optimized with games in mind while DirectX is specialized toward the consumer market (games).
If you want to play the latest and most "advanced" games and at the "highest quality", it is undisputed that you get a PC built for that purpose. And even at "mid level", you would be able to pay half what you paid for a Mac bootcamp. Either a Mac or PC can play games that a "budget computer" can play at acceptable performance. But only die hard MacTards that have no clue what they are talking about would deny that a PC is the choice for anything above that requirement.
A true/false view of GL's level of optimization would neglect to account for the improvements seen on OS X in just the last few months.
http://barefeats.com/nehal18.html
http://barefeats.com/nehal18.html
Yes, and I think the gains in Portal are even higher than reported by barefeats. Something went wrong with their Portal test on their Mac Pro, since they got better framerates on an iMac at equivalent settings. Macbenchmark.com reports 2x improvements on Portal with the 4870 at high settings.