ag_6502 soft core with phase-level accuracy

Great project!

I am reading the source code of genstates. but I cannot figure out how the optimization work. Does it use Quine–McCluskey algorithm?

Best regards.

Chou from Japan.

Yes, my resulting algorithm seems to be very close to Quine-McCluskey algorithm. The main difference (as I can imagine) between my algorithm and Quine-McCluskey's is that in my program optimization is done recursively, so it is possible to decrease the number of multiplications with the use of braces. Maximum depth of the recursion is specified by -O parameter.

I observed a very strange behaviour of Xilinx Verilog optimizer with different values of the -O parameter in genstates. In a simple program the best (the most compact) results are achieved without recursive optimization (-O1, default) what is equivalent to the use of Quine-McCluskey algorithm, but complex design (such as a design of a microprocessor) requires infinite level of optimization (-O-1). But in other cases one may use other intermediate levels of optimization which leads to different maximum levels of the recursion.

In general, increasing the level of the recursion, you will get more compact but more slow design because it will decrease the parallelism of the logics.

Thanks for your replying!