If you follow fast blockchains like Solana or Aptos, you probably know how annoying network traffic jams get. When heaps of people try to trade at the exact same time, network nodes struggle to process transactions smoothly. I came across a study about a system called Omakase that aims to fix this exact bottleneck. It stops computer cores from sitting idle while keeping heavy traffic under control.
Researchers wanted to see if blockchains with multiple proposers could run transactions faster without crashing under heavy conflict. They built Omakase, a framework where computers use idle processing cores to pre-execute proposed blocks and map out data conflicts ahead of time. A greedy scheduler then combines these conflict maps across a whole round of blocks using a balance heuristic. When tested on 15,000 Ethereum mainnet blocks with around 2.4M transactions, Omakase reached a 2.47x speedup over sequential execution. The system matched rival throughput while staying markedly more stable during high traffic spikes. On top of that, overlapping the integration work with execution hid 94% of its measured pipeline cost.
If you want the full details, the paper is called Omakase: Execution Congestion Control through Conflict-Graph Integration in Multi-Proposer Blockchain and it was written by Dakai Kang, Ray Neiheiser and Alberto Sonnino, with several coauthors. You can find it at arxiv.org/abs/2611.17184.
This is not financial advice. The information provided is for educational and informational purposes only.
The cover image was generated by AI.