[Libre-soc-bugs] [Bug 770] Discussion and Finalisation of Which Cryptographic Primitives to Implement

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=770

--- Comment #13 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Jacob Lifshay from comment #11)
> i assumed that you would infer I meant using instruction count and probable
> implementation strategy as a proxy for speed.

no, as it is industry-standard practice to differentiate explicitly
between performance (usually Dhrystone MIPS) and performance/watt.

> > SV being completely abstract and an architecturally independent
> > ISA the only thing we can possibly claim right across the board
> 
> then don't claim right across the board,

i had to do the analysis for the Due Diligence, so i know that it is.

> have several). maybe claim that it doesn't have terrible performance issues
> for smaller microarchitectures if svp64 is implemented in hardware (e.g.
> excluding trap & emulate)

these still reduce instruction count, and performance is the least concern,
which is precisely why i made it clear that it is imperative *not* to focus
on performance.

also the implementor is free and clear to implement precisely and exactly
that hardware desired for the specific (sole) application for which that
silicon is exclusively dedicated.  if they fail to add the required
instructions
and damage performance as a direct result that is their problem to fix.

right across *all* of these options reduction in instruction count is the
sole constant that reduces power consumption (except where implementors
fail to do a decent job, which is, again, out of scope)

summary: performance claims it is imperative that we make none WHATSOEVER,
the only claim we can reasonably make is "performance/watt reduction"
and steer well clear of stating what that is, because it is too implementation-
specific.  we have to *have* an implementation in order to measure it, and
we are about 2 years away from that.

so please: *do not* make performance claims, do not make proposals containing
performance claims, do not make presentations showing performance claims,
do not publicly state anywhere in any discussions that performance is claimed
to be greater. you will be required to provide proof, and get us into
trouble for being unable to back up the claim, or, worse, commit us to
needing to provide proof of that claim, when we have zero funding to be
able to do so.

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