I'm writing to draw fellow Debian ARM users' attention to the
Parallella project currently on Kickstarter. Link:
http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone
On most extant ARM-based SoCs/dev.boards, the coprocessor (i.e. GPU)
is encumbered by NDAs, closed source drivers, binary firmware,
undocumented instruction sets, etc.
Although I'm fan of Parallela, just a small comment on GPUs vs. Parallela: although both are computing technologies exploiting parallelism, they are also quite different beasts. GPUs tend to be vectorized, often leaning towards VLIW architecture and have considerable amount of application-specific units and instructions. Parallela, in contrast, is more like a grid of interconnected NUMA/shared-memory integer/float scalar 32-bit rather traditional RISC microcontrollers. Both approaches have their strengths and weaknesses...