Intel Graphics Hit High Point With Linux 3.6 Kernel

[phoronix]

Phoronix: Intel Graphics Hit High Point With Linux 3.6 Kernel

Testing of the latest Linux 3.6 kernel that's presently under development has revealed some additional OpenGL performance improvements with Intel graphics, at least concerning the latest-generation "Ivy Bridge" processors.

http://www.phoronix.com/vr.php?view=17761

[uid313]

OpenArena performance

Why is the FPS so much lower in OpenArena 0.8.8 than in 0.8.5?

[archibald]

Why is the FPS so much lower in OpenArena 0.8.8 than in 0.8.5?

It has more advanced shaders and sparkly things (can you tell that I'm not au fait with the nitty gritty details?).

[curaga]

Michael, you should mention what is "gen4" in normal people speak. Not everyone has an Intel decoder chart tattooed across their chest

[archibald]

Michael, you should mention what is "gen4" in normal people speak. Not everyone has an Intel decoder chart tattooed across their chest

I believe the GMA 4500 is the start of Gen4 though I did need to spend a while trawling wikipedia to find this out :-)

[imamdxl]

Still slower than Windows driver

http://openbenchmarking.org/prospect...cef5388eaab19f

this is pretty upsetting even with open source drivers and with all the collaborations, performance can not be excelled against Windows driver.

Windows driver does best what it supports even with fewer OpenGL extensions as well as almost no bugs when running games (even they use hecks to satisfy the specs but users look for end results).

Where Linux driver adding more and more extensions, slower, continuesly breaking one game or another while improving

I can not play Amnesia the Dark Descent anymore with Mesa 8, 8.1, git. The last Mesa for running Amnesia the Dark Descent properly was 7.11.2.

Features like point sprites is broken for gen4 from the beginning of Mesa and still not fixed

[Veerappan]

Still slower than Windows driver

http://openbenchmarking.org/prospect...cef5388eaab19f

this is pretty upsetting even with open source drivers and with all the collaborations, performance can not be excelled against Windows driver.

If you take a look at the description and system specs in the linked results, you'll see that the comparison you're looking at is with the 3.4.0 kernel and is 3 months old. Given the amount of improvement between the 3.4 and 3.6 kernels, the Linux performance might be equal to, or higher than, the windows performance on those machines now for those chips.

[roughl]

Still slower than Windows driver

http://openbenchmarking.org/prospect...cef5388eaab19f

this is pretty upsetting even with open source drivers and with all the collaborations, performance can not be excelled against Windows driver.

Its not slower in all cases: http://openbenchmarking.org/prospect...fc89be48b2e792

[roughl]

Still slower than Windows driver

http://openbenchmarking.org/prospect...cef5388eaab19f

this is pretty upsetting even with open source drivers and with all the collaborations, performance can not be excelled against Windows driver.

Its not slower in every benchmark: http://openbenchmarking.org/prospect...fc89be48b2e792

[orome]

If you take a look at the description and system specs in the linked results, you'll see that the comparison you're looking at is with the 3.4.0 kernel and is 3 months old. Given the amount of improvement between the 3.4 and 3.6 kernels, the Linux performance might be equal to, or higher than, the windows performance on those machines now for those chips.

take lightsmark test on i7:
improvement from 3.4 to 3.6 in Micheal's test is 42->57 FPS (1.35x)
interpolating the linked results would give 77.3FPS vs. 85.1FPS (i7 Linux vs. Windows)

still lower, but pretty close. so kudoz to intel developers