HaRoLd
HaRoLd
With the LongsightF gadget this seems like less of a priority as the serialised circuit is quite small compared to the full SHA256 merkle tree proof gadget.
I'm not sure if Bazel would be useful, CMake does a good job and I've never used Bazel before. However, last time I said that I was using automake/autotools and...
Using the new montgomery form multiplier this can be reduced to 2 constraints per bit. One thing remaining is to convert the Y coordinate of the resulting point to its...
Optimised pedersen hashes have now been implemented: * https://github.com/HarryR/ethsnarks/blob/master/src/jubjub/pedersen_hash.cpp This can be adopted to be used with the merkle tree: * https://github.com/HarryR/ethsnarks/blob/master/src/gadgets/merkle_tree.cpp * https://github.com/HarryR/ethsnarks/blob/master/ethsnarks/merkletree.py
I think it could be split into the following sub-repos: * `ethsnarks-python` - Python support * `ethsnarks-js` - Javascript support (hopefully) * `ethsnarks-cxx` - Gadgets and components in C++ for...
From: https://github.com/matterinc/sapling-crypto/blob/master/src/alt_babyjubjub/mod.rs#L5 * scaling = 1911982854305225074381251344103329931637610209014896889891168275855466657090 * a' = 21888242871839275222246405745257275088548364400416034343698204186575808495616 == -1 = a*scale^2 mod P * d' = 12181644023421730124874158521699555681764249180949974110617291017600649128846 == -(168696/168700) = d*scale^2
https://github.com/AztecProtocol/barretenberg looks promising too
Interestingly, GMP is *very slow* in Emscripten with WebAssembly, and 30% slower than an optimised version of a 256bit multiply operation. Improving performance on WebAssembly would require investigating field operations...
https://ethresear.ch/t/double-batched-merkle-log-accumulator/571 includes some interesting ideas for a batched incremental merkle tree
> If we had a keccak256 gadget, then computing intermediate hashes along the merkle tree would be cheaper AFAIK, as long as the source isn't completely outdated. The on-chain cost...