[Libre-soc-dev] v3.1B prefix

[Alexandre Oliva]

On Dec  5, 2020, Luke Kenneth Casson Leighton <lkcl at lkcl.net> wrote:

> On 12/5/20, Alexandre Oliva <oliva at gnu.org> wrote:
>> As in, does it still make any sense for me to get back to the
>> compression estimator, which I was going to return to as soon as I was
>> done with email?

> yes funnily enough it does because the 16 bit ISA still needs
> statistical analysis for forming SV-P32.

'k, I've just pushed what I managed to implement this weekend.
It's most of the 10- and 16-bit opcodes, but not (yet):

- 2.1.4 conditional branches (b and bl are done)
- 2.1.9 condition register (nothing done here)
- 2.1.10 system (nothing done here)

there are a number of '???' annotations that you might want to look at.
Also a number of commented-in assumptions within the definition of
copcond, that it would be good to double-check.

There are a lot of encoding rules that are quite specific to the
compressed-mode proposed encoding, so I'm not sure how useful they will
be for analysis of different encodings.  The work to model the current
rules has been quite involved, though it's grown easier in the end, as
primitives and accessors were already in place.

Even with all these opcodes put in, the compression rate for gcc's text
segment was less than 3%.  It's not looking incredibly compelling :-(

-- 
Alexandre Oliva, happy hacker  https://FSFLA.org/blogs/lxo/
   Free Software Activist         GNU Toolchain Engineer
        Vim, Vi, Voltei pro Emacs -- GNUlius Caesar