[libre-riscv-dev] ASIC layout questions
Luke Kenneth Casson Leighton
lkcl at lkcl.net
Thu Jun 25 18:10:17 BST 2020
On Thu, Jun 25, 2020 at 6:02 PM <whygee at f-cpu.org> wrote:
> If the timing is too tight for the data signal to go all the way
> through the array of latches, you can AND the unary read and write
> signals to trigger a MUX at the output (which I suppose you did in
> the wrapper).
yes, basically, that's what i did. with all of the register
"addresses" being unary, it's easy to do. for the binary versions
it's harder to do because you have to decode the binary reg numbers to
unary first *or* perform an O(N*M) array of binary-address
comparators, N=read_ports, M=write_ports.
More information about the libre-riscv-dev