Intel X.Org 2.12 Driver To Render Text/Glyphs Faster

[phoronix]

Phoronix: Intel X.Org 2.12 Driver To Render Text/Glyphs Faster

Intel's next quarterly Linux graphics driver update will be out around July and while nothing too exciting has yet to emerge within the X.Org DDX driver that will be released as xf86-video-intel 2.12, there is one interesting merge today. This next Intel X.Org driver update should offer significantly faster text / glyphs performance...

http://www.phoronix.com/vr.php?view=ODI4MA

[FunkyRider]

When I see the title, I was more sort of hoping for a 2~10x improvement. 1.08~1.34 on low end CPU means even lower improvement in mid-hi CPU systems, which might average to a 1.1x improvement, which does not worth the effort put in. IMO

[Chad Page]

Haven't looked at the code, but since it's probably GPU (bandwidth?) bottlenecked I could see even Sandy Bridge gaining nicely from this

In any case, 10% faster text is nice. 2-10x would be much more meaningful on something really slow, like the Neomagic driver (the bane of many old P2/P3 laptops under Linux...)

[brent]

The log shows quite a few performance improvements and fixes for corruption bugs. Chris Wilson is doing a good job, or so it seems.

[aliendude5300]

Yay...? It's still not that good, considering I get over 1000kglyphs minimum on both tests, using the open source ATI driver. Glad to see Intel working on the driver though!

[squirrl]

By the time we get full support for intel chips they will be obsolete.

[FunkyRider]

Yay...? It's still not that good, considering I get over 1000kglyphs minimum on both tests, using the open source ATI driver. Glad to see Intel working on the driver though!

You have to know that the test object in the article uses an ATOM cpu. What kind of CPU do you have for your test? I believe a proper Core 2 with Intel graphics should score way higher than 600k/s

[Craig73]

I for one appreciate Chris Wilson's optimizations. There is a lot of Intel hardware out there and any performance improvement keeps existing hardware viable for longer.

Yes 3D optimizations would be nice, but 2D is still quite significant and if the cost/benefit ration is good then tune away!!!