[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 #14 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
i apologise i hit "back button" and lost a reply.

battery low so must be brief.

greatly appreciate the review and insights on this exceptionally complex
deceptive code.

let us however focus, please do comment on this logical reasoning for any flaws
or misassertions

* early analysis shows SRNOR and SRNAND to be functionally equivalent even down
to the "unknown" transitions
* the difference is the inversion of inputs and outputs, which, given boolean
algebra manipulation, makes sense.
* a careful code-morph can put SRNAND in place of SRNOR
* yosys appears to have $sr and _SR_* however they are undocumented and need
investigation
* nmigen limitations can be worked around by using cocotb and "specials"

basically what i am saying is that if you were to focus on adding SRNAND to the
cell library, for the March deadline, we can use it.

again, to emphasise: someone reading and asking questions on this exceptionally
difficult design is genuinely appreciated.

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