[Libre-soc-dev] NLnet Ongoing Grant

[Luke Kenneth Casson Leighton]

brief, as Andrey will be doing a complete write-up of changes
made to bug 961.

Jacob came up with an extremely important use-case example:
reverse-gear on Data-Dependent Fail-First on sv cmpi.

as SVP64 work (spec, simulator, insndb, binutils, tests)
is absolute absolute top priority over and above everything
else without fail without exception the tough decision had
to be made to get this in, with an appropriate budget.

that means that the optimisation to create an in-order core
has to go, as a lower-priority goal. we have TestIssuer:
it does the job.

a reminder to everyone as there are still questions when i have
made it very clear many times: the absolute top priority is
completing SVP64 and the Simulator and associated insndb and
binutils work.

future work - libopid for example - is *future* work for
which appropriate costings and budget allocations will be
required.  my estimations put full conversion of the entire
Libre-SOC codebase (8 affected repositories) in the ballpark
of EUR 25,000.  FP HDL we had Jacob's estimates at around
EUR 24,000.  those two alone make up an entire new NLnet Grant.

which we are not going to get if the *outstanding* work, using
the *existing* codebase, is not completed to 100%.

Jacob's getting on well with pow(x,y,mod) with an implementation
of Knuth Algorithm D.

Sadoon's making good incremental progress on poly1305.

l.




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