Hey guys. As i've stated in titled, i want to connect 4 evaluation boards in order to achieve parallel processing, for let's say array multiplication. The master device should distribute the array elements on each of the slaves, which will excecute the multiplications for a specific range and return the results to the master. Is this achievable? If so, what is the easiest and fastest communication protocol to achieve this ?
If they are evaluation boards, your choices seem to be ethernet, CAN, or serial via the PMOD connector. I think if you look around, you find that there are many math kernels (libraries) that support SMP architectures for most LA operations - BLAS and LAPACK being very popular, thought on Intel architectures.
The S7G2 is also a Cortex M4 architecture itself - the "4" implying that it contains a DSP core tht can perform math on single, double, or quad values simultaneously. Take a look - http://www.disca.upv.es/aperles/arm_cortex_m3/curset/CMSIS/Documentation/Core/html/group__intrinsic___s_i_m_d__gr.html
Good luck on your project.