bazza参与话题 2消息总数 2

bazza 2026-06-27 21:02:48 ↗ Hyperliquid 用户询问订单修改延迟能否降至500ms以下

What's the realistic mainnet round-trip to modify a resting order (send → ack)? seeing ~880ms, is sub-500ms achievable?

What's the realistic mainnet round-trip to modify a resting order (send → ack)? seeing ~880ms, is sub-500ms achievable?

bazza 2026-06-27 05:55:07 ↗ Hyperliquid 用户测量 Hyperliquid 订单修改延迟高达 880ms

Quick question for anyone running latency-sensitive execution on mainnet: what's a realistic round-trip to modify/replace a resting order — request sent to accepted-ack received? I'm measuring ~880ms+ per op, which lines up with the Glassnode numbers and looks like mostly engine processing, not network. Three things I'm trying to figure out: 1. Does a single modify (or batched cancel+place) come back faster than sequential cancel-then-place, or roughly the same? 2. Is ~880ms the real floor on current mainnet, or are people getting sub-500ms acks with priority fees / colocation / running a node? 3. Do maker strategies here actually run sub-second reprice loops, or is the norm to quote wider and reprice less because the round-trip makes tight repricing impractical?

Quick question for anyone running latency-sensitive execution on mainnet: what's a realistic round-trip to modify/replace a resting order — request sent to accepted-ack received? I'm measuring ~880ms+ per op, which lines up with the Glassnode numbers and looks like mostly engine processing, not network. Three things I'm trying to figure out: 1. Does a single modify (or batched cancel+place) come back faster than sequential cancel-then-place, or roughly the same? 2. Is ~880ms the real floor on current mainnet, or are people getting sub-500ms acks with priority fees / colocation / running a node? 3. Do maker strategies here actually run sub-second reprice loops, or is the norm to quote wider and reprice less because the round-trip makes tight repricing impractical?