[Libre-soc-bugs] [Bug 1155] O(n^2) multiplication REMAP mode(s)

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=1155

--- Comment #53 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Jacob Lifshay from comment #52)
> (In reply to Luke Kenneth Casson Leighton from comment #50)
> > therefore the design of svshape3 and 4 need really to be
> > done in conjunction with the ed25519 and/or pow-mod to
> > get a clear picture / use-case.
> 
> ok, I'll work on trying to fit them into the bigint mul and powmod
> algorithms soon, probably tomorrow.

awesome. i recommend doing individual unit tests first
(test_caller_svp64_bigmul.py or test_caller_svp64_svshape.py)
working up to
that vertical-first loop idea, keep it to a real short
loop of say 2x2 or 3x2 64-bit otherwise it takes too long
to run. we need *small* unit tests, running for
*no more than 30 seconds total*, first. these act as "demos"
and end up with significant comments (see strncpy in ldst test
for what i am looking for)

write and use svshape4 initially as that's the one that can
have (needs) "explicit" control (targetting individual SVSHAPEs
but also including an offset).

remapyield.py does now have modulo, triangle, and rhombus capability,
you'll need to tie it in to ISACaller just like the other remaps.
set mode=0b11 rather than mode=0b00

once we know how many svshape4s are needed, plus an svremap instruction,
*then* we have enough proper working konwledge to design svshape3.

to give you some idea, this is the schedule for ed25519
https://bugs.libre-soc.org/show_bug.cgi?id=773#c3

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