The GPU Tech Conference deadline is fast approaching on November 3,
2011 23:59 PDT:
http://www.gputechconf.com/page/call-for-sessions.html
I would like to see some CLPP representation even if it is just at the
level of a poster.
Many topics we could address such as:
1) GPU/APU/CPU performance differences.
2) Finding the cutoff on various hardware platforms where OpenCL
overhead and memory transfer costs make it worth calling the OpenCL
version of a sort or scan. Especially with relation to the C++ STL.
3) Major hangups we are finding between OpenCL devices that are
causing portability issues.
My focus right now is mainly on optimizing CLPP for the AMD A8-3800
(Llano) APU. If there isn't much interest in collaboration I will
probably just submit a poster, but I would be happy to integrate into
a larger CLPP paper.