[Openicc] GSoC & oyranos

[Kai-Uwe Behrmann]

Am 31.03.09, 11:27 +0200 schrieb Cyrille Berger:
> On Tuesday 31 March 2009, Kai-Uwe Behrmann wrote:
>> Am 27.03.09, 19:31 +0200 schrieb Yiannis Belias:
>>> I've been hearing lately about OpenCL. If there is a way to test it in
>>> linux before the end of summer(who knows...), I want to give it a try.
>>
>> Agreed, OpenCL is a interessting option and will be even better portable
>> than GLSL based plug-ins.
> I agree, but I have doubts to see OpenCL on linux before the end of the year.
> It might comes in proprietary drivers some time this year, but I would still
> only expect them to be available in fall or winter. And for open source
> drivers, we might get some developement version, but I have doubt we will get
> something usefull (even in developement version) before next year (see [1] for
> latest update, but note that both phoronix and Zack Rusin are usually over
> optimistic on schedules/features).
>
> That said, I trully believe OpenCL is definitively the way to go over GLSL.

It seems that a future OpenCL implementation would be based on the C++ 
llvm framework. So my main reservation seems pointless with llvm appearing 
virtualy everywhere (Mesa/Xorg). This was probably the intention not to 
write OpenGTL in project idea page. So I have a few question around 
OpenGTL.

CTL apepars to be a established language for our main field of 
application - colour transformations.
How about Shiva? Is it fully compatible with PixelBender alias Hydra? I 
found no specification for the format used by PixelBender. It seems to 
change from time to time and Adobe is free to do so.

Can Shiva and CTL programms be automatically converted to OpenCL to get 
advantage of its expected hardware support in LLVM [2]? This would relax 
the need to care about OpenCL from user side - Krita, Oyranos ...

In OpenCL pointers are much more restrictively used than in traditional C.
Would this be an issue for a LLVM based convertor? Can or must CTL and 
Shiva programms be optimised for OpenCL? I read LLVM has to convert to IL 
anyway. But optimisation is not always an easy task.

How does you solve multi channel processing in OpenGTL if at all? Does you 
simply split processing for non channel crossing operators? Or ist this to 
be done by the user?

What are your plans for OpenGTL? Do you plan a transition to OpenCL when 
it will be available? Your suggestion to Yiannis above make me thinking 
this is likely.


best regards
Kai-Uwe Behrmann
-- 
developing for colour management 
www.behrmann.name + www.oyranos.org

[1] http://www.opengtl.org/
[2] http://llvm.org/