SNA Continues To Be Far Better Than UXA, GLAMOR

[smitty3268]

Because it runs on top of OpenGL

Well, at least you got 1 part right.

and OpenGL is not meant to decode video.

It has nothing to do with decoding video.

It's a hack by AMD

Glamor is an Intel project.

because they don't want to expose the real thing(TM) in their GPUs

2D accel hardware no longer exists in new GPUs.

because of patent issues.

Glamor is used to reduce code duplication, nothing to do with IP or any patents.


Also, note that it actually outperforms UXA according to these benchmarks. That means it likely outperforms EXA as well (which is what the radeon drivers used to use). It's just the highly complicated Intel-specific SNA that does extremely well here.

[mrugiero]

Expect that is used for cairo-image here. Something which OpenGL should be able to do very well?

It is still an extra layer, against directly calling the hardware.

[tstrunk]

How do all you fare with graphics corruption when using SNA?
Since enabling it, me and a few of my colleagues using Ivy Bridges all have some visual artifacts especially in normal desktop applications (Bad redrawing while scrolling in Chrome, black boxes while scrolling in firefox, etc.)
Everything is quite stable and there are no crashes, but the "experience" is still better with UXA.

[JS987]

How do all you fare with graphics corruption when using SNA?

2.99.903 works fine for me. You can report bugs at https://bugs.freedesktop.org/

[ickle]

How do all you fare with graphics corruption when using SNA?
Since enabling it, me and a few of my colleagues using Ivy Bridges all have some visual artifacts especially in normal desktop applications (Bad redrawing while scrolling in Chrome, black boxes while scrolling in firefox, etc.)
Everything is quite stable and there are no crashes, but the "experience" is still better with UXA.

I'm sorry to hear that, unfortunately it seems to have found both of the minor bugs on Ivybridge. The first appears to be an issue with LLC invalidation. Progress is going slow as I am reliant on users to do my testing for me. The second is a hardware bug that requires a workaround. Please never hesitate in reporting issues upstream.

[tstrunk]

I'm sorry to hear that, unfortunately it seems to have found both of the minor bugs on Ivybridge. The first appears to be an issue with LLC invalidation. Progress is going slow as I am reliant on users to do my testing for me. The second is a hardware bug that requires a workaround. Please never hesitate in reporting issues upstream.

Thanks! Those seem to be exactly the bugs I'm experiencing. From now on I will file them upstream.

[Drago]

How do all you fare with graphics corruption when using SNA?
Since enabling it, me and a few of my colleagues using Ivy Bridges all have some visual artifacts especially in normal desktop applications (Bad redrawing while scrolling in Chrome, black boxes while scrolling in firefox, etc.)
Everything is quite stable and there are no crashes, but the "experience" is still better with UXA.

I also see graphics corruption. Thankfully it is a rare thing, I could live with.

[darkbasic]

Please never hesitate in reporting issues upstream.

Does it apply for Glamor bugs too? Becuase I have a serious issue with glamor on AMD+KDE which I can probably reproduce with Ivy Bridge. Hopefully Intel will be more interested in fixing it?

[ickle]

Does it apply for Glamor bugs too? Becuase I have a serious issue with glamor on AMD+KDE which I can probably reproduce with Ivy Bridge. Hopefully Intel will be more interested in fixing it?

Eric might as he seems to have a renewed interest in Glamor. But if the bug is never reported anywhere the chances of it ever being fixed are severly diminished...

[fscan]

Doesn't SNA work under the assumption that it has an integrated gfx card with shared memory? If yes, these numbers don't say much in the dedicated gfx card world.
Are there benchmarks comparing GLAMOR to EXA/UXA on radeon?