Hyperliquid Consensus, Order Execution, Latency, and Throughput

Summary

This documentation overview explains the trading architecture of Hyperliquid. It describes HyperBFT, a HotStuff variant, as the consensus mechanism, with validators producing blocks in proportion to native tokens staked to them. The state is divided between HyperCore, which contains margin and matching engine state, and the general-purpose HyperEVM. The exchange says HyperCore uses on-chain order books and a single transaction order established through consensus.

The page reports end-to-end latency for a geographically co-located client as a 0.2-second median and 0.9-second 99th percentile, and approximate mainnet capacity of 200,000 orders per second. It attributes the current throughput limit to execution and says consensus and networking can scale further once execution improves. These figures are platform claims in an overview, not independent measurements; latency depends on client location, and the page gives no benchmarking methodology, trading-cost analysis, or evidence about strategy profitability.

Key ideas

• HyperBFT provides consensus and transaction ordering for Hyperliquid.
• HyperCore holds margin and matching engine state, while HyperEVM provides general-purpose execution.
• The overview reports sub-second latency figures for a co-located client and approximate capacity of 200,000 orders per second.
• Execution is identified as the throughput bottleneck, and the reported platform figures lack independent validation in this document.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.