[Libre-soc-dev] NLnet cryotoprimitives grant approved
luke.leighton at gmail.com
Tue Aug 10 13:47:24 BST 2021
with many thanks to NLnet, the EUR 50,000 grant to research and develop Draft cryptographic primitives and instructions to the newly-open Power ISA has been approved.
unlike RISC-V where full transparency and trust is problematic and there are many participants whose interests may not necessarily align, the OpenPOWER initiative, which has been in careful planning for nearly 10 years, is a much less crowded space and, crucially, does not require non-transparent membership of OPF in order to submit ISA RFCs (Requests for Change)
[non-OPF members cannot participate in actual ISA WG meetings and certainly cannot vote on RFCs, but they can at least submit them. whereas whilst the RISC-V Foundation's Commercial Confidence Requirements are perfectly reasonable, the blanket secrecy even for submitting RFCs is not]
we at Libre-SOC aim to use this process, based on taking apart key strategic cryptographic algorithms back to their mathematical roots, then applying Vector ISA design analysis and seeing what can be created.
examples include going back to the fundamental basis of Rijndael, and instead of creating hardcoded custom silicon for MixColumns as is the "normal" practice, adding a generic Galois Field ALU and a generic Matrix Multiply system. another is to design instructions suitable for "big integer math"
this in turn means that the resultant ISA would be ideally suited to the experimental development of future cryptographic algorithms for use in securing wallets and other purposes related to blockchain management.
[as bitcoin stands we cannot possibly hope to compete with custom silicon dedicated to SHA hash production, however we would very much like to see a future version of bitcoin that uses far less power yet retains its high strategic value, and, at the same time, like e.g. monero RandomX, is better suited to a general-purpose Vector Supercomputer ISA, which is what we are developing]
OpenPOWER's commitment to a transparent RFC process allows us to do that without compromising trust: no discussions that we participate in will ever be behind closed doors.
if anyone would be interested to participate or collaborate on this, we have funding available, and welcome involvement in designing and testing an ISA suitable for securing bitcoin for end-users in a fully transparent fashion.
More information about the Libre-soc-dev