RadeonSI Compute Shader Patches Revised

[phoronix]

Phoronix: RadeonSI Compute Shader Patches Revised

For those anxious to see compute shaders for then having OpenGL 4.3 support by the RadeonSI Gallium3D driver for GCN GPUs, the latest patches have been published...

http://www.phoronix.com/scan.php?page=news_item&px=RadeonSI-Compute-Shaders-V2

[microcode]

Perhaps some day soon they can completely kill Catalyst on Linux. Seems like RadeonSI is approaching comparable API completeness, comparable performance, and better compatibility.

Exciting times.

[Ericg]

Perhaps some day soon they can completely kill Catalyst on Linux. Seems like RadeonSI is approaching comparable API completeness, comparable performance, and better compatibility.

Exciting times.

Unlikely-- As noted in the original AMDGPU docs, Catalyst has some rather nasty performance hacks in it that the mesa guys would never accept, they also get used to validate a specific version against scientific applications that need absolute correctness.

[asdfblah]

Kinda off-topic; this looks insteresting, to say the least: https://lists.freedesktop.org/archiv...il/112385.html

[boxie]

Unlikely-- As noted in the original AMDGPU docs, Catalyst has some rather nasty performance hacks in it that the mesa guys would never accept, they also get used to validate a specific version against scientific applications that need absolute correctness.

I am quite looking forward to the 16.04 upgrade cycle because of AMDGPU finally being "ready" for my Tonga card.

[bridgman]

The primary reason for keeping the Catalyst GL driver is support for compatibility profiles - there are no plans to ever add that to Mesa.

That (no compatibility profile support in Mesa) is a good decision IMO, but not all market segments agree with me yet

[Amarildo]

Unlikely-- As noted in the original AMDGPU docs, Catalyst has some rather nasty performance hacks in it that the mesa guys would never accept

But that wouldn't stop the Mesa guys from creating things from scratch that fall on their criteria of acceptable, right?

IMO, Mesa is almost there, we can ALMOST kill Catalyst. There is, however, a very nasty flickering problem that AMD aparently can't/won't fix.

[bridgman]

IMO, Mesa is almost there, we can ALMOST kill Catalyst. There is, however, a very nasty flickering problem that AMD aparently can't/won't fix.

Amarildo, which flickering problem are you talking about ? Is there a ticket open for it ?

I found one that was closed as a Steam overlay issue, and another on Pitcairn that seems to be fixed by reverting a specific quirk.

Thanks !

[Amarildo]

Amarildo, which flickering problem are you talking about ? Is there a ticket open for it ?

I found one that was closed as a Steam overlay issue, and another on Pitcairn that seems to be fixed by reverting a specific quirk.

Thanks !

Thanks for responding

This: https://www.phoronix.com/forums/foru...-i-report-this

I tried using the same modeset as the one used by Catalyst, but it didn't solve the problem.

I haven't opened a ticked because I don't know where the problem is located.

[eydee]

The primary reason for keeping the Catalyst GL driver is support for compatibility profiles - there are no plans to ever add that to Mesa.

That (no compatibility profile support in Mesa) is a good decision IMO, but not all market segments agree with me yet

You mean that radeon pro hybrid driver, right? It was already stated (by yourself, if I'm not mistaken) that Catalyst will never get xorg 1.18 support. This means that in a few months no supported linux distribution will exist that can run Catalyst. Eventually even Debian and the likes will upgrade xorg. Basically Catalyst (fglrx) is dead.