[Libre-soc-dev] gcc binutils sv cryptoprimitives etc

Luke Kenneth Casson Leighton lkcl at lkcl.net
Wed Jan 20 08:55:57 GMT 2021

jacob you make some good points however are still missing the point and the
primary objective.

the primary objective - if i can word this "badly" - is to use cryptography
as an "excuse" to get certain bitmanipulation instructions funded,
peripherals funded (nmigen RGMII, nmigen USB-ULPI), skywater layout and PDK
funded, all of which are missing.

if the end result is cryptographic algorithm implementations that need
further funding to get upstream *we have still met our primary objectives*
and provided the basics for *other people* to improve on.

we do not need excuses, "upstream will reject this or that".  this is not
helpful, at all.

we are focussing on the low level.

if upstream rejects patches needed to provide security because we came up
with a different paradigm to solve constant time that's *not our problem*.

if there are cache timing attacks that is the focus of a completely new
research grant.  not our problem.  someone else can do a PhD to fix that,
get lots of kudos for doing so.

i keep telling you again and again we are under enormous time crunch and
have to be extremely draconian in what we choose to focus on.

once we have the basics other people will come forward and help out.

if however we try to do everything ourselves by increasing the scope we
will run out of time and money.

am i making this clear enough?

we cannot ask NLnet for ten times the money needed to do ten times the
actual work needed.

if we severely and DRASTICALLY cut the scope to what IS achievable within a
given grant budget it provides the basis for **LATER** work.

given how badly constant time will f*** our ability to complete any work in
reasobable time it *HAS* to be declared out of scope.

if once we have the funding someone cones forward and says, "you're doing
it wrong, i can do better and i will help you meet the core critical
primary objectives" GREAT.

if not, so what. we still met the primary objectives, we have a working
proof of concept that allows VCs to gain confidence in the overall SV

is this now clear what the objectives are?

can i get some help on what's needed rather than what's not needed, and
what i need rather quickly is a list of crypto primitives similar to GF
arithmetic that are part of the underlying mathematics of blockchain,
hashing algorithms and cryptography.

the obvious other one is big integer math (carry across vectors).

are there any other fundamentals common to SHA-NNN i do not mean the VSX
instructions i mean the underlying mathematical primitives.

etc etc.


crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68

More information about the Libre-soc-dev mailing list