One of the touted uses of the Zynq is as an accelerator of software computation. When I attended X-fest in May, an entire session was devoted to this subject. However its treatment on this forum has been sparse at best.
I have gone thru ZynqGeek's tutorial for create a custom peripheral on the AXI bus. I can successfully write and read from the peripheral registers from an ARM elf file. However I have been less successful in extending the user logic to make use of the periperal registers to do parallel computation in custom HDL. When I attempt to include a post-translate simulation verified module (and sub-modules) into the user_logic(.v) module, it does not work. XST ends up trimming the submodule because I obviously have not done it properly.
The tutorial that one post mentioned by Silica is outdated and only adds an output port to be fed to external pins. What I am interested in is as was discussed in the zynq acceleration session.
Do you have any reference designs that include such a custom peripheral for which a coprocessor for, say, a dantziq simplex linear programming matrix solution could be implemented on the ZedBoard. I have initially chosen AXI bus slave registers as the means to supply the M by N matrix, but realize that using AXI DMA and/or the ACP may be more efficient.
This area of PS/PL collaberation for advanced embedded designs seems an appropriate step for advancing the knowledge base here. Can we expect reference designs of this nature?
