All of your ideas are great, specially the OpenGL output op, which isn't a lot of work and should give good performance (mem. transference between the cpu and gpu is very expensive).
But for now, what would be great is people testing the code and performance comparisons/profiling to know what can be improved, we don't need GPUs to do that, OpenCL should run on CPUs :)
On Tue, Apr 3, 2012 at 7:34 AM, Jon Nordby <jononor@xxxxxxxxx> wrote:
glad to see that GPGPU/openCL support has landed in GEGL master. What
are the plans for future work in this area?
Obviously we want more and more operations to be implemented and
optimized for openCL, but have you given any thought to the following
* Having an openGL output operation, which will render the output
already present as an openCL image object to a openGL texture. The
application can then use this (either directly or through one of the
UI integration libraries) to display the composition.
* Color management of viewport using CL/GL. Let the above operation
(or a separate operation) accept a LUT specifying the device profile
to use for the render and do the necessary conversions.
* GPU powered export/import (JPEG/PNG). These algorithms are also
computationally expensive and could benefit from being GPU
Jon Nordby - www.jonnor.com
_______________________________________________ gegl-developer-list mailing list gegl-developer-list@xxxxxxxxx http://mail.gnome.org/mailman/listinfo/gegl-developer-list