Re: [OT] Byte-wise boolean optimisation

Thomas Pircher wrote:
Hi all,

I'm wondering if there exist algorithms for byte-wise boolean optimisation,
much like the Quine-McCluskey works for a single bit.

Basically I have 8 truth-functions f7(.) ... f0(.) representing the bits
7..0 of the result byte, and I would like to find an optimised function
f(.) which gives the same result as the single-bit-functions together and
uses 8-bit parallel boolean operators.

Obviously I can write
f(.) = f7(.) << 7 | f6(.) << 6 | ... | f0(.)
but this is not very elegant. Is this a common problem that someone has
solved in a better way than my straight-forward (naive) attempt?

I'm unclear - do you want a one byte answer, from an 8 byte variable set, and you want to use common byte-wide operators, to get to the answer ?
Sounds a fairly big ask, as you will need 'leave alone' operations many times.

You could easily feed the eqns into any CPLD/FPGA tool flow,
and inspect the output eqns for each bit, which will of course be
optimised, but using bit operators. Then, you could work backwards, to
see if any can be merged into multi-bit operations.
Atmels WinCUPL has choices of optimise.

-jg

.