Performance

TL;DR. In March 2026 I published a README table saying my Rust as-of join, FlowState, was 1.8 times faster than Polars (10 ms against 18 ms)1. When I wrote a reproducible benchmark in October, the result reversed: in my runs on one Apple M4 laptop FlowState’s median was 13.3 ms against 9.9 ms for Polars on 1M left rows23, and 194.9 ms against 135.8 ms on 10M left rows45. I removed the March claim. Rebuilding the March kernel did not bring the lead back6, and I do not know why Polars is faster or what produced the original number. ...

This post predates my current standard for evidence (every number traced to a captured run or a cited source). Treat figures here as unverified. The Problem You’ve optimized your multicast application. The code is clean. The algorithm is sound. Yet your latency numbers lag behind competitors, and you can’t figure out why. The bottleneck isn’t in your logic—it’s in the cable running under the floor. This isn’t flashy. No one writes blog posts about fiber optics at conferences. But here’s the reality: in high-frequency trading, real-time video delivery, and large-scale sensor networks, the choice between a generic fiber cable and an optimized one can mean the difference between 500ns and 5µs latency. That’s a 10x spread, and it starts at the physical layer. ...