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

[tstellar]

As I understood it, that is why AMD decided to go with VLIW4, because DX10 shaders, couldn't fill five slots of VLIW5.
Tom, is lib/CodeGen/DFAPacketizer is what you are talking for? I don't see any concrete Target references here.

DFAPacketizer is the correct file. I just pushed the VLIW work I've done so far to the r600-vliw branch here: http://cgit.freedesktop.org/~tstellar/llvm/ If you take a look a the lib/Target/AMDIL/R600Packetizer.cpp file it should give you a good idea of how it works.

[Drago]

DFAPacketizer is the correct file. I just pushed the VLIW work I've done so far to the r600-vliw branch here: http://cgit.freedesktop.org/~tstellar/llvm/ If you take a look a the lib/Target/AMDIL/R600Packetizer.cpp file it should give you a good idea of how it works.

I will give it a look, during the holidays.
You should write posts here more often, so gpu driver developers wannabes like me, to understand more easily the big picture. Or maybe a blog or something. GPUs are not simple devices so this would help for community recruitment.
Happy holidays, Tom.

[Drago]

DFAPacketizer is the correct file. I just pushed the VLIW work I've done so far to the r600-vliw branch here: http://cgit.freedesktop.org/~tstellar/llvm/ If you take a look a the lib/Target/AMDIL/R600Packetizer.cpp file it should give you a good idea of how it works.

Tom, file AMDGPUTargetMachine.h, probably has merge conflict in it, and there is no *.inc files in AMDIL target, except SIRegisterGetHWRegNum.inc. Are these going to be pushed?

[Asariati]

You should write posts here more often, so gpu driver developers wannabes like me, to understand more easily the big picture. Or maybe a blog or something. GPUs are not simple devices so this would help for community recruitment.

I second this request!

[bridgman]

FYI Tom does have a blog at :

http://tstellar.blogspot.ca/

[Drago]

Tom, file AMDGPUTargetMachine.h, probably has merge conflict in it

Merge conflict fixed in subsequent commit. My bad.

[tstellar]

I will give it a look, during the holidays.
You should write posts here more often, so gpu driver developers wannabes like me, to understand more easily the big picture. Or maybe a blog or something. GPUs are not simple devices so this would help for community recruitment.
Happy holidays, Tom.

As John noted, I do have a blog, but I don't post very often. Maybe I can change that in the future. If you have questions you can always stop by the irc channels #radeon or #dri-devel on irc.freenode.net and ask them there.

[tstellar]

Tom, file AMDGPUTargetMachine.h, probably has merge conflict in it, and there is no *.inc files in AMDIL target, except SIRegisterGetHWRegNum.inc. Are these going to be pushed?

The .inc files are generated by llvm-tablegen when you run make, but that branch I pushed up was really just a code example. I haven't tested it, and I'm not even sure if it will build.

[t.s.]

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

Still a minority. But not a small number, isn't it? 10% income is not small amount.

[Death Knight]

Still a minority. But not a small number, isn't it? 10% income is not small amount.

Means, AMD's profit from linux users are ~%10 (or much). Questions is, if AMD spend 1/9 of the "windows driver development fund" to "linux driver development"?
I think they do not. If this is true, means, they doesn't pay back to us but they are paying back to windows users with the profits from linux users GPU purchases...