There is a chance if you are able to run Pharo on that platform and support python wrappers for CUDA (python is already support for CUDA for regular nvidia GPUs but I dont know about this platform ) you could use my Atlas lib to access Cuda from Pharo. Atlas is a library I am making that allows you to use Python libraries from Pharo and its code I use for my project Ephestos to control Blender the 3d app. ����

Not as cool or powerful / flexible as having an actual Pharo wrapper but it could be a good replacement till such library arrives.��

There is also Pharo wrappers for OpenCL by another Pharo developer , if this platform support OpenCL (it seems that it currently it does not) it may be possible to modify the wrappers to support this platform as well.��


On Wed, Aug 27, 2014 at 11:29 PM, volkert@nivoba.de <volkert@nivoba.de> wrote:
Wouldn't this a nice platform for Pharo? I mean a Pharo with full CUDA and OpenGL Support ... i would like it. :-)

https://developer.nvidia.com/jetson-tk1
http://devblogs.nvidia.com/parallelforall/jetson-tk1-mobile-embedded-supercomputer-cuda-everywhere/
http://elinux.org/Jetson_TK1
http://developer.download.nvidia.com/embedded/jetson/TK1/docs/Jetson_platform_brief_May2014.pdf
https://www.youtube.com/watch?v=9teSvCL0NX0
https://www.youtube.com/watch?v=YfUHei3M0Jw

BW,
Volkert

--
www.nivoba.de

"The more complex an object, the larger the investment in learning to use it, and the greater the resistance to abandon it.", NW