|
| 1 | +--- |
| 2 | +title: Weekly Summary – March 10, 2025 |
| 3 | +authors: |
| 4 | +- will |
| 5 | +tags: [progress, update, weekly] |
| 6 | +--- |
| 7 | + |
| 8 | +This week saw significant progress in simulation capabilities with the successful comparison of Rust and Haskell simulations across **[90 scenarios](https://github.com/input-output-hk/ouroboros-leios/blob/weekly-update-2025-03-10/analysis/sims/2025w11/analysis.ipynb)**. We conducted our **first mainnet-scale [analysis](https://github.com/input-output-hk/ouroboros-leios/blob/weekly-update-2025-03-10/analysis/sims/2025w11xl/analysis.ipynb) of Leios** on a [mainnet-scale network](https://github.com/input-output-hk/ouroboros-leios/blob/leios-2025w11/sim-rs/test_data/realistic.yaml) of 3000 nodes, revealing unexpected performance benefits from network topology. **[Sharding performance analysis](https://github.com/input-output-hk/ouroboros-leios/blob/weekly-update-2025-03-10/analysis/shard-performance.ipynb)** provided important insights into optimization strategies. Both simulation implementations received improvements to enhance realism and comparability, while our formal methods team developed initial trace verification tools for Short Leios. |
| 9 | + |
| 10 | +## Simulation comparison |
| 11 | + |
| 12 | +- Compared [90 scenarios](https://github.com/input-output-hk/ouroboros-leios/blob/weekly-update-2025-03-10/analysis/sims/2025w11/analysis.ipynb) between Rust and Haskell simulations at tag [`leios-2025w11`](https://github.com/input-output-hk/ouroboros-leios/releases/tag/leios-2025w11) |
| 13 | +- Recent fixes and adjustments enabled meaningful comparison between simulations |
| 14 | +- Identified [issues](https://github.com/input-output-hk/ouroboros-leios/issues?q=is%3Aissue%20state%3Aopen%20label%3Aquestion) requiring further investigation. |
| 15 | + |
| 16 | +### Analysis of mainnet-scale simulation |
| 17 | + |
| 18 | +- Completed first [analysis](https://github.com/input-output-hk/ouroboros-leios/blob/weekly-update-2025-03-10/analysis/sims/2025w11xl/analysis.ipynb) of Leios on a mainnet-scale network simulation using Rust simulator |
| 19 | +- Discovered 3000-node mainnet-scale network transports IBs faster than artificial 100-node network |
| 20 | +- Identified "shortcut" edges in larger networks as likely reason for improved transport speed. |
| 21 | + |
| 22 | + |
| 23 | + |
| 24 | +### Performance analysis of sharding |
| 25 | + |
| 26 | +- Created [computational models](https://github.com/input-output-hk/ouroboros-leios/blob/weekly-update-2025-03-10/analysis/shard-performance.ipynb) showing relationship between fraction of shards without an IB and expected number of extra IBs |
| 27 | +- Evaluated performance characteristics of the simplest sharding scheme. |
| 28 | + |
| 29 | + |
| 30 | + |
| 31 | +## Haskell simulation |
| 32 | + |
| 33 | +- Fixed bug in Relay protocol preventing full diffusion of votes |
| 34 | +- Changed priority of certified EB to be included in RB |
| 35 | +- Added support for output log format sharing a common subset with the Rust sim |
| 36 | +- Analyzed TCP realism in comparison to idealized diffusion |
| 37 | + - Found higher IB rate and size leads to better diffusion times |
| 38 | + - Identified ledger state access as significant source of latency. |
| 39 | + |
| 40 | +## Rust simulation |
| 41 | + |
| 42 | +- Added more information to logs (total size of IB, parent id of RB) |
| 43 | +- Implemented same EB selection strategy as Haskell |
| 44 | +- Added validation of IB headers before propagation to neighbors |
| 45 | +- Investigating why the simulation doesn't encounter as much congestion as Haskell. |
| 46 | + |
| 47 | +## Formal methods |
| 48 | + |
| 49 | +- Developed initial trace verifier for Short Leios simulation traces in `leios-trace-verifier`. |
0 commit comments