Hello,

the reason I asked about IM on GPUs is
that GPUs can process graphic object much
faster (10-30 times) and that the acceleration
rate of GPU development is faster than the
known Moore cycles (doubling in 6 month
instead 18). This is because many graphic
algorithms can be parallelized and the
GPUs are (massive) parallel architectures
whereas many transistors in CPUs are
occupied in pre- and post processing
operations for sequential processing.

But not only graphical applications
benefit from GPUs because they are become
programmable and a scientific community
is working on general purpose GPU programming
see the overview on
http://graphics.idav.ucdavis.edu/publications/print_pub?pub_id=907
or the website http://www.gpgpu.org

GPU programming is slowly moving from an academic
playground to the mainstream since big players realize the
opportunities:
-AMD/ATI,  NVidia and third parties are offering boards
and development tools
- Apple uses in OSX 10.4 Core Image that can
use a GPU for near real-time image and video processing
http://developer.apple.com/macosx/coreimage.html
it uses with the CIKernel language a subset of
OpenGL that is relevant for 2D graphic
-Google buying Peakstream


Nevertheless general purpose GPU programming is
different (Stream Processing) and there are no standards
at the moment only OpenGL for the graphic side.
But there are approaches to make GPU programming
easier and Google supports a lot of Open Source projects
so perhaps they will releasing some Peakstream technology
for the public after they achieved their in-house goals.


It would be big step for IM if it could use a GPU
especially for batch processing where nothing is
displayed and the GPU is doing practically nothing.


regards
Günter















_______________________________________________
Magick-users mailing list
[email protected]
http://studio.imagemagick.org/mailman/listinfo/magick-users

Reply via email to