Jon Phillips wrote: > I don't understand everything you are saying, but it is great to see > the details coming out on why Milkymist is amazing. Great work!
Heh, I'd say it's not so much about M1 per se but about who really owns your FPGA. The presentation does a very nice job of dispelling various myths and misconceptions surrounding logic synthesis (and P&R), and of showing how the seemingly overwhelmingly complex task can be broken down into more manageable smaller problems. This reminds me a bit of the front cover of the Green Dragon Book: http://ecx.images-amazon.com/images/I/51EQPxCipwL._SL500_.jpg There, the knight attacks the dragon of "Complexity of Compiler Design" with a set of tools ("LALR Parser Generator" on the lance, "Syntax Directed Translation" on the shield, and "Data Flow Analysis" on his steed's armor). I also like the flow of the presentation. Each time you're about to raise an objection, the next slide has the answer or at least gives you a starting point. Regarding implications on marketing of the M1, the work on LLHDL itself doesn't do much to support any claims of greatness of the M1 in its primary field of application, i.e., VJing. However, I think we could position the M1 as one reference platform for people interested in exploring the depths of synthesis with LLHDL. Considering the possibility of damaging an FPGA with a sufficiently unfortunate mis-configuration, a low-cost board may be needed to complement the M1 as an experimental platform. This could be either something coming from the M1/qi-hw universe or a third party product. E.g., a simple M1 daughter board with a CPLD (CPLDs are the "little brothers" of FPGAs) or a low-end FPGA could also be useful as a programmable test pattern generator or dummy device for analyzing FPGA behaviour. - Werner _______________________________________________ http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org IRC: #milkymist@Freenode
