First steps on AMD Fusion

With the arrival of the Fusion platform we decided to tune the code generation of Minotaurus to support AMD GPUs as well. Introducing basic structured OpenCL code generation, the latest AMD driver accepts and runs Minotaurus on Fusion. The speedup is currently suboptimal, the memory management will be tuned also to the special case of Fusion. It can perform now ~30 fps with 512x512 resolution on Zacate. Of course, hand optimized code could do it better, but this is still generated code from C++, using templates and other candies. OpenMP parallelized version of the same code can achieve ~8 fps on the double core Bobcat.