AMDGPU Open-Source Driver Code Continues Maturing

[marek]

Also set "force_glsl_version=0" for Heaven in the environment.

[marek]

Do you (AMD devs) think to upstream all the parts in time for kernel 4.2 and the next mesa?

Yes. After the kernel merge window opens, we'll try to merge everything.

[frosth]

Someone earlier was asking for dpm status, any progress with Rx 285? I know Alex could tell more, but i didn't see him here.

Also set "force_glsl_version=0" for Heaven in the environment.

verde, i used "MESA_GL_VERSION_OVERRIDE=4.1 MESA_GLSL_VERSION_OVERRIDE=410|" and it worked for tessmark (tesselation-may) and tessmark&heaven (tesselation-mar[ch]).
I try 'may2' later.

[mao_dze_dun]

I still think what AMD are doing is very crappy. They should at least help with the creation of a proper GUI to work with the open source driver. Otherwise it sure feels like I'm being punished for buying a 290x instead of the *cough*cheaper*cough* 285.

[bridgman]

I still think what AMD are doing is very crappy. They should at least help with the creation of a proper GUI to work with the open source driver. Otherwise it sure feels like I'm being punished for buying a 290x instead of the *cough*cheaper*cough* 285.

Sorry, I don't understand the point you're trying to make. Are you saying there is a GUI available for 285 (presumably via amdgpu) which is not available for 290X ? Or are you talking about Catalyst (which is available for both cards) ?

[niner]

No. Why would you want that? There is practically no difference other than the fact amdgpu is planned to be used by Catalyst.

Because people hope that if the driver is used by Catalyst, more man power will be available to fix issues. For example, I'd really love to have working power management on my R9 270X before it's time to upgrade to a newer GPU again. Having to reboot just to change the power profile just sucks. Yes, it's reported as https://bugs.freedesktop.org/show_bug.cgi?id=77394 and yes, I'd even try to fix it myself. But not even questions about which component the error would likely be in are answered :/

[jrch2k8]

Yes. After the kernel merge window opens, we'll try to merge everything.

Marek i tested your swapbuffers-offload branch and seems stable(tf2, witcher2, Civ5, Gallium nine games, ut4 demos, unigine demos rock solid) but it break openmax compilation, other than that is peachy. just informing. this week im testing the tessalation may2 branch on TAHITI too

Awesome work, TyVM. BTW marek is your egl 1.5 branch safe for test? i mean is know to break wayland or compositors? if not i can test that one too

[duby229]

I still think what AMD are doing is very crappy. They should at least help with the creation of a proper GUI to work with the open source driver. Otherwise it sure feels like I'm being punished for buying a 290x instead of the *cough*cheaper*cough* 285.

It depends on what you mean by GUI. The OSS drivers do have some pretty awesome graphical tools. The stats overlay is great, and it can technically be considered graphical

[dungeon]

What is "DCC"?

To me main new feature of GCN 1.2 chips - delta color compression. It is not essential to have and dunno how much will perfromance be affected without it.

[GreatEmerald]

Those jsrcipt requrements for quoting/posting/etc.. are awful, so i guess people often posting several times in a row because of that, so do i

No, it's because it looks like multiquoting is broken no matter the JavaScript settings... I get a "Post Reply (N)" button at the top there, but all it does is load the last post quoted, not all of them. Quite annoying...