Intel OpenGL Performance: OS X vs. Windows vs. Linux

[phoronix]

Phoronix: Intel OpenGL Performance: OS X vs. Windows vs. Linux

As mentioned last week when publishing the OS X 10.8 vs. Ubuntu Linux benchmarks, a large Intel OpenGL driver performance comparison was being carried out at Phoronix. The comparison is now compete and here are the results when comparing the Intel HD OpenGL graphics performance under Apple OS X 10.8, Microsoft Windows 7 Pro, and Ubuntu Linux 12.04/12.10. The results of this Intel OpenGL gaming performance comparison are quite interesting, but reveal some troubling Linux facts.

http://www.phoronix.com/vr.php?view=17818

[TeoLinuX]

So far for the many praises on Intel about their friendliness on open source.
They have the tech and SW know-how to make their HW go 50% faster and openGL 4.0 compliant, but they use it only on Windows...

Maybe I'm a little bit harsh, and Ok, great job to the OTC, but just like for AMD: couldn't the OSS-driver team be helped internally more?
IP to defend? on AMD side maybe, but come on, on HD3000....

Anyway, it slowly and constantly progresses. At least there's a hope , unlike the perspectives on future we had 5 years ago.
And the collaboration with Valve should pay

[rudl]

wow linux driver is faster than i thought.

[Kano]

Ok, intel can run some engines which are 10y old but it would be much better when it could run idtech 5 (even when done via wine). But thats not the case. Tests would be more interesting with hd 4000 however, then you could at least compare heaven.

[Anarchy]

With all the money Apple has I thought their drivers will completely obliterate everything. It appears that their drivers are not perfect at all (compared to Win7), and often are slower than Linux.

[Spectre]

I remember reading about 'Link-Time Optimising' when compiling the Linux kernel. I know it's not fully ready yet but might that in the future increase performance when it comes to OpenGL?

[schmidtbag]

Windows 7 did perform the best overall but I wouldn't say it destroyed linux. Overall I saw a pretty health balance between all 3 OSes, if you choose the highest-performing Ubuntu tests. Each OS had its strengths, which I personally found weird considering how many of these games are based on the same engine.

[FourDMusic]

This could actually be viewed as a win for linux. While we still can't compete against the windows graphics stack, I feel that we have surpassed OS X - at least with the intel driver. For many of the benchmarks, we tie or surpass the OS X performance. There are a few cases where linux does quite a bit worse (Xonotic, high resolution issue, but linux handles high quality better... interesting).

The fact of the matter is that OS X is optimized for a specific piece of hardware whereas a distribution like Ubuntu strives to be as generic as possible so that it will work on a wide range of hardware. Maybe, if you completely optimized the linux operating system and graphic stack for a specific set of hardware, than these number would be different. How different, I don't know (though I'll try and find out - at least for my hardware).

[curaga]

I remember reading about 'Link-Time Optimising' when compiling the Linux kernel. I know it's not fully ready yet but might that in the future increase performance when it comes to OpenGL?

No Intel GPU is CPU limited in anything more than the old OA. So no, LTO won't likely have much effect on intel fps.

[gamerk2]

So far for the many praises on Intel about their friendliness on open source.
They have the tech and SW know-how to make their HW go 50% faster and openGL 4.0 compliant, but they use it only on Windows...

Maybe I'm a little bit harsh, and Ok, great job to the OTC, but just like for AMD: couldn't the OSS-driver team be helped internally more?
IP to defend? on AMD side maybe, but come on, on HD3000....

Anyway, it slowly and constantly progresses. At least there's a hope , unlike the perspectives on future we had 5 years ago.
And the collaboration with Valve should pay

Playing devils advocate: Why the automatic assumption its the driver that's at fault? I mean, theres a LOT of drivers out there with reduced functionality compared to their Windows counterparts. Maybe the issue is less with the driver writers and more with the host OS?