Does anyone know when OpenSource ATI GPUs power options are fixed?

[bridgman]

Personally I am happy with my E-350, PM is little worse than in windows, but I could use some hw video decoding, and 3D performance. AMD please, optimize your open shader compiler.

Are you following the work Tom has been doing with llvm ?

[crazycheese]

One thing also needed is an open source version of the Catalyst Control Center for the free driver. Changing power settings using cat and echo is so last century. Something imaginative like Radeon Control Center.

Driconf ?
Although if reclocking is integrated 2gether with SLI(dreams, dreams that are on the market for tens of years), it will be much more than "dri"-conf.

[bridgman]

Does that mean that you have made public all the power management information? Because i have been hearing promises about documentation and information about powermanagement, i think, from the very start of the radeonhd project, which will soon be 5 years ago.

For this specific issue (making dynpm work on a broader range of systems) yes, in the sense that the hardware programming is already done and the problems are related to *when* that programming is done. There are basic PM functions common across multiple generations and that information has been in use for a couple of years -- other functions are specific to single generations of GPU and are not documented yet; some are used, others are not, and we're looking at which ones are worth documenting/supporting.

Plus, the boring bits 80% of the time needs to be done by paid developers, the chances of finding someone interesting in doing boring bits from the community are very slim, i believe that we stated something along those lines in our proposal to AMD.

Yeah, it may be that simple, at least for PM. I don't remember seeing that in the proposal but it's good advice anyways.

Oh, and when did ATI last provide documentation for the boring bits? All docs being made available since the death of RadeonHD, afaik, are the ISAs, and those are made available from AMDs GPGPU department (if that still exists), not by the former ATI.

The existing documentation is still relevant right up to NI, and the ISA docs from the GPGPU group (former ATI btw) cover most of the shader core differences between generations. My understanding is that even with the ISA docs the information on doing attribute interpolation in the shaders is still a bit thin, so we need to go back and document some more there. SI (aka GCN) is really where we need the next big slug of documentation, and that is being worked on now.

[Drago]

Are you following the work Tom has been doing with llvm ?

Yes, I am watching Tom's patches, but I believe his work is more focused to compute on r600g, and compute/graphics on radeonsi.
I wish more optimal graphics performance on r600g.

[curaga]

One thing also needed is an open source version of the Catalyst Control Center for the free driver. Changing power settings using cat and echo is so last century. Something imaginative like Radeon Control Center.

I would be happy with just non-debug, stable places in /sys. Currently AFAIK some non-debug things like checking your clocks still require debugfs.

[t.s.]

Are you interested in selling cards for Linux desktop market or not.

Harsh, but good question. 'Cause it will determine my next buy, as in near future, me and friend wanna buy new comp.

Oh, and.. 1 thing that these manufacturer doesn't seems understand (or forget, maybe?) is: LINUX {geek} here, although minority, have many people asked them for recommendation about hardware (or laptop) they ought to buy. So, you can said that 1 geek here not just represent him/herself, but, say, represent at minimum 10 person. Translate to: when these manufacturer dissapointed their LINUX {geek}, they lost 10 other at minimum..

But then again, I think they can't see it this way

[darkbasic]

That means 10% instead of 1%, we are still a minority

[PsynoKhi0]

Every single time the usual nay-sayers show up in an AMD-related thread...

[schnelle]

One thing also needed is an open source version of the Catalyst Control Center for the free driver. Changing power settings using cat and echo is so last century. Something imaginative like Radeon Control Center.

+1

We really need more user friendly way to change power settings until dynpm becomes reality.

[tstellar]

Yes, I am watching Tom's patches, but I believe his work is more focused to compute on r600g, and compute/graphics on radeonsi.
I wish more optimal graphics performance on r600g.

The r600 LLVM backend can be used for graphics too, and it comes fairly close to matching the current r600 compiler as far as number of piglit test passes. It is just missing support for some of the texture instructions and a few other things. It should produce much better code than the current r600g compiler once it makes use of the LLVM VLIW packetizer, which is something that I'm hoping will be done as part of the Open Source compute effort. Adding support for the texture instructions shouldn't be too difficult of a task. It is not a priority for me at the moment, but someone else from the community could easily do this if they were interested.