Luke Kenneth Casson Leighton
lkcl at lkcl.net
Wed Dec 16 19:25:29 GMT 2020
On Wed, Dec 16, 2020 at 6:45 PM Jacob Lifshay <programmerjake at gmail.com> wrote:
> Umm, actually MM increments fastest --
now, you see: i couldn't tell that, and because the table is laid out
2 dimensionally, and is so large, i couldn't deduce it either.
a *linear* table on the other hand, just like you've laid out, *now*
that's much more obvious. it'll be long enough that i'd recommend it
be done as a separate page (one each for INT/FP and one for CR). with
- and this is important - the SRAM index number as one of the columns.
also what would be very handy to have in that table would be the
envisioned "prefix" needed (2 and 3 bit versions) to turn the
"standard" field (5-bit RA, RS in the case of INT, FRA for FP, and
3-bit BA in the case of CRs *and* we also need one for the 5-bit BFA
as well) into SV names.
actually now i think about it, it's probably a good idea to write a
small python program that auto-generates the table(s). now *that*
would be clear.
More information about the Libre-soc-dev