[libre-riscv-dev] [Bug 154] Cell for Dependency Matrices is needed

[bugzilla-daemon at libre-riscv.org]

http://bugs.libre-riscv.org/show_bug.cgi?id=154

--- Comment #2 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
here is a comparison between the truth tables of SR NOR and SR NAND

SR NOR

S R Q
0 0 Latch
0 1 0
1 0 1
1 1 0

SR NAND

S R Q
0 0 *INVALID STATE*
0 1 1
1 0 0
1 1 Latch

as i suspected there is an inversion of both the output and of the inputs.  in
many cases both the Q and ~Q are needed, so the negation of output is not
costing gates.

however the inversion of the inputs, and in particular the instability, will
both need investigating.

of course normally a full DFF would be deployed to prevent that instability
however that is a cost of 10 gates vs 2 which is where the estimated 5 fold
increase from 50,000 to 250,000 gates just for one of the DMs comes from.

-- 
You are receiving this mail because:
You are on the CC list for the bug.