Other Linux Things I Learned From AMD At GDC 2014

[Vim_User]

Given the way that open source gfx performance is catching up with Catalyst on your hardware (arguably better already for 2D and video), is the logic here that you would be punishing us for supporting you with open source drivers rather than closed source ?

I would rather see the logic in: They deprecated the proprietary driver long before offering an alternative with similar performance and features in his case, in my case (RS780) even without offering a driver with similar features (any updates on the UVD front?). Again a marvelous example of incompetent marketing on AMD's site, they should have waited with deprecating the driver at least for another or better 1.5 years.

Anyways, bridgman, you may have more insight on that, maybe you can answer that question: How plan the few developers that seem to work on Catalyst for Linux to make the switch from fglrx to radeon if they don't even have the manpower for proper changelogs (something that could be generated automatically anyways by the version control system)?

[bridgman]

I would rather see the logic in: They deprecated the proprietary driver long before offering an alternative with similar performance and features in his case, in my case (RS780) even without offering a driver with similar features (any updates on the UVD front?). Again a marvelous example of incompetent marketing on AMD's site, they should have waited with deprecating the driver at least for another or better 1.5 years.

Remember that the decision to remove support for GPU generations from the "main" Catalyst driver is primarily driven by Windows requirements but since the code is shared across OSes the Linux support gets hit as well.

Anyways, bridgman, you may have more insight on that, maybe you can answer that question: How plan the few developers that seem to work on Catalyst for Linux to make the switch from fglrx to radeon if they don't even have the manpower for proper changelogs (something that could be generated automatically anyways by the version control system)?

Again, shared code between multiple OSes makes release note generation non-trivial. Some bugs fixed in the common code affect all OSes, others do not -- and in cases where fixes affect Linux there may be no app that uses the affected area so claiming an improvement would be bogus. Some of the bug fixes are for workstation customers with proprietary apps, and we often aren't supposed to talk about those at all.

Not saying it can't be done but it's really not as simple as it might first seem.

[johnc]

Linux systems have to dismiss obsolete grpahical server xorg replacing it with modern solutions as wayland and mir.

Bite your tongue!

X.org is amazing... bordering on spectacular.

[Azrael5]

Bite your tongue!

X.org is amazing... bordering on spectacular.

Search online and learn why wayland is superior to Xorg, so will be you to bite yourself.

[BSDude]

Search online and learn why wayland is superior to Xorg, so will be you to bite yourself.

You can't get the sarcasm without it being written in bold, can you?

[Azrael5]

You can't get the sarcasm without it being written in bold, can you?

If you read sarcasm in my message, means that sarcasm is evident. Do you consider that user as an idiot?

http://www.phoronix.com/scan.php?page=article&item=x_wayland_situation&num=1

[mannerov]

Remember that the decision to remove support for GPU generations from the "main" Catalyst driver is primarily driven by Windows requirements but since the code is shared across OSes the Linux support gets hit as well.



Again, shared code between multiple OSes makes release note generation non-trivial. Some bugs fixed in the common code affect all OSes, others do not -- and in cases where fixes affect Linux there may be no app that uses the affected area so claiming an improvement would be bogus. Some of the bug fixes are for workstation customers with proprietary apps, and we often aren't supposed to talk about those at all.

Not saying it can't be done but it's really not as simple as it might first seem.

Do you know if the plan is to drop the proprietary DDX too?

I imagine that if catalyst uses libdrm and the radeon kernel module, then Mesa and Catalyst could live together.
If Catalyst doesn't implement gbm for example, we could use Mesa if we need gbm (for example that would allow to use glamor for xwayland).

I have a laptop with enduro, and I used Catalyst for a moment last year.
I wanted to be able to use the dedicate gpu when I wanted, but not suffer from power consumption issues.
I had a lot of issues installing (as with my previous experiments with Catalyst).
When it finally worked, I was satisfied enough of the result to keep it for several months.
However some things didn't satisfy me:
. Boot was 55 seconds without Catalyst, and 1min40 with it.
. If I changed something to kernel, Xorg or the intel DDX, or for some Catalyst updates, I would have a lot of issues.
. There was no vsync support when using the dedicated card, and there was really a lot of tearings.

If the attempt to use the radeon kernel module and libdrm succeed, the two first issues would be solved, and last one would be bypassed because I think it would just drop the current mode 'full integrated card or full dedicated card' to an optimus/enduro like solution, like we are to see with Mesa.

I switched to Mesa when dpm was ready. But if Catalyst dramatically improves, I would use it again to play games needing high OpenGL support.

[Vim_User]

Remember that the decision to remove support for GPU generations from the "main" Catalyst driver is primarily driven by Windows requirements but since the code is shared across OSes the Linux support gets hit as well.

Which basically comes down to: We don't care if that decision impacts Linux, you are second class customers. Way to go.

Again, shared code between multiple OSes makes release note generation non-trivial. Some bugs fixed in the common code affect all OSes, others do not -- and in cases where fixes affect Linux there may be no app that uses the affected area so claiming an improvement would be bogus. Some of the bug fixes are for workstation customers with proprietary apps, and we often aren't supposed to talk about those at all.

Non-disclosure parts for workstations aside, a changelog does not have to claim improvements, it has to document changes.

But you avoided the real question: UVD is low priority due to scarce resources, Mantle is low priority due to scarce resources, changelogs are low priority due to scarce resources, so where do the developers take the time from for such major changes?

[GreatEmerald]

You can't get the sarcasm without it being written in bold, can you?

Coming from johnc, I'm pretty sure it wasn't sarcasm (note his user title).

Which basically comes down to: We don't care if that decision impacts Linux, you are second class customers. Way to go.

Well, saying anything else would simply be dishonest.

[Awesomeness]

Who cares about potatoes that can't even do 1080p?

So far 10 million people care: http://www.vgchartz.com/

Also Benalib claimed that no gamer uses AMD hardware. So far 10 million console gamers use AMD APUs. More gamers use AMD GPUs because GameCube, Wii, Wii U, and Xbox 360 also have AMD GPUs (with IBM CPUs) which results in over 200 million console gamers alone. PC gamers are not even included in the statistics.

My post was solely meant to disprove him (which I did). It was not meant to argue about your personal interests.