[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 #4 from Staf Verhaegen <staf at fibraservi.eu> ---
Some of my comments:
- A SR latch could be easily added to the standard cell library and I will do
that. I'm not sure though current nmigen + yosys will actually infer SR
latches.
- If one wants to use SR latches nmigen is not the right language. nmigen is
based on normal latches with just a reset signal which is only an init value. I
do think it is better to rephrase the code as logic with one sync statement of
one wants to keep on using nmigen for the scoreboards. Certainly on nmigen
level one should not try to take into account the undefined behaviour of a
latch when both inputs are 1 or 0. That should be taken care of behind the
scene by nmigen and yosys.
- Efficiency of simulation should not be the guidance on how to implement
things. Better to switch to cocotb + verilator or other solution for simulation
then.

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