Quadrvium
@qdrvm_io
Followers
95
Following
16
Media
22
Statuses
116
Blockchain infrastructure developers. Contributing to @polkadot, @libp2p, and @ethereum
Joined August 2023
QLEAN – a lean Ethereum consensus client written in C++ by Quadrivium team – achieved multi-client interoperability with with Ream and Zeam implementations, together reaching 3SF-mini consensus 🚀
1/ 🎉 PQDevnet0 is complete! The 1st integration milestone in bringing post-quantum cryptography to Ethereum. Three clients from @ReamLabs , @zeamETH, and @qdrvm_io achieved finality with 3SF-mini! Read the full story: https://t.co/03uOStq0uA Quick summary 👇
0
1
2
11/ We consider our simulator to be a useful tool to check how practical different networking and compute assumptions may be for the post-quantum transition. Feel free to use it to test your assumptions! The full report explains how to reproduce our result
0
0
1
10/ Using the Aggregators-only Grid, 200Mbps bandwidth, and doubling the recursive SNARK rate to 20/sec, we achieved a final aggregation time of just 2.3 seconds
1
0
1
9/ We pushed the configuration values to find a viable path to aggregate much faster, leaving more of slot time for block and global SNARK propagations
1
0
1
8/ With all factors considered (8192 signatures, 128KB SNARKs, 1000sigs/second, 10snarks/second aggregation rates), here's the time to final aggregation with 100Mbps bandwidth: * Standard Gossipsub: 4 seconds * Grid: 2.3 seconds
1
0
2
7/ One approach is to expose only p2p addresses of public aggregators, while keeping attesters addresses more private, by letting them send their signatures directly to aggregators
1
0
1
8/ One area in which Grid falls short compared to Gossipsub is privacy, as it exposes the P2P addresses of validators. The report discusses how to mitigate some privacy concerns
1
0
1
7/ Another nice property we observed in our simulations is the scaling of the Grid topology with increased bandwidth. The Grid's aggregation speed improves more significantly as bandwidth increases.
1
0
1
6/ To solve this, we tested a structured Grid topology. Unlike Gossipsub, the Grid is deterministic, ensuring any message reaches any peer in just 2 hops.
1
0
2
5/ Our simulations show the current Gossipsub approach is too slow. Existing EIP-7870 50Mbps bandwidth, the combination of network latency and compute delay pushes aggregation time past the 4-second slot limit
1
0
1
4/ We modeled aggregation speeds of 1000 signatures/sec into the first layer of SNARKs, and then 10 SNARKs/sec for the final global proof
1
0
2
3/ We simulated this scenario at scale with 8,192 validators. The problem is twofold: * Can the network handle the abundance of signature data? * Can we generate the recursive SNARKs fast enough?
1
0
2
2/ On top of that, we're replacing small BLS aggregations with SNARKs. Each SNARK proof is 128KB, adding another massive data-sharing challenge for the P2P layer
1
0
1
1/ The challenge begins with data size of signatures. The PQ signatures we simulated are 3,072 bytes which is much larger to BLS signatures that are used for attesting Ethereum blocks today
1
0
2
🧵 We're excited to share our report on simulation of Post-Quantum (PQ) signature aggregation! https://t.co/NWE6CzXMd7 With support from an @ethereumfndn grant we investigated requirements needed to aggregate thousands of signatures within a four-seconds lean consensus slot
hackmd.io
We aim to gossip and aggregate num_signatures post-quantum signatures (each of size signature_size) across the network within a time limit max_time_seconds.
1
3
19
Ref. 1689 for boosting development of Polkadot-SDK got approved! Thank you to everyone who supported us!
* Faster DA and erasure coding * Single-honest collator parachains * Compact and fast archive storage * Efficient state sync * Deterministic integration tests These are the improvements that we are introducing to @polkadot . Consider supporting us by voting in the referendum!
0
0
3
* Faster DA and erasure coding * Single-honest collator parachains * Compact and fast archive storage * Efficient state sync * Deterministic integration tests These are the improvements that we are introducing to @polkadot . Consider supporting us by voting in the referendum!
$288,000 USDC Treasury Spend Proposal: Ref. 1689: Quadrivium Polkadot-SDK improvements proposal 1 Feat. @qdrvm_io
2
2
10
For our purposes we needed only QUIC transport. So external contributions are welcome to add transports and protocols like TCP, WS, Noise, Yamux. Once the new library matches the feature set of the old library it will be replacing the original cpp-libp2p library
0
0
0
The original library was originally developed in 2019 and since then was battle-tested in many @Polkadot networks. Now in 2025 C++ has evolved. Crucially now it supports coroutines, so writing async code for protocols implementation is getting much easier. New library takes
1
0
0
We are publishing a new version of C++ @libp2p library. https://t.co/biYQTG6eAF It is written using modern C++ with features like coroutines and concepts. The code is now much simpler to use. The outstanding feature is C wrappers, so library could be also used to develop
github.com
C++23 implementation of libp2p with C bindings. Contribute to qdrvm/leanp2p development by creating an account on GitHub.
1
2
4