Compiler-Assisted Floating-Point Error Analysis and Profiling with FPChecker

A half-day tutorial in Hamburg is pitching FPChecker as the gadget that helps developers catch sneaky math mistakes before those tiny errors snowball into broken results. In plain English: it watches programs doing decimal-heavy calculations, flags weird stuff like impossible numbers, and helps people see where accuracy goes off the rails. There’s even a hands-on angle with laptop demos, downloadable tools, and backup AWS cloud machines for anyone whose setup refuses to cooperate. Very practical, very serious — and then the comments arrive.

Because the real action is in the peanut gallery. One of the loudest reactions wasn’t about the tool at all, but a classic internet demand: "Is there actually video of the talk?" In other words, no matter how useful the tutorial sounds, part of the crowd basically wants proof of life and a replay link. Another commenter gave FPChecker a thumbs-up but immediately turned into a product manager, wishing the graph used base 2 so it matched how computer numbers are actually stored. Translation: cool tool, but the chart is bugging me.

And then came the spicy academic side-eye. One commenter was surprised there was no mention of Herbie, another tool in the same orbit, and wondered whether the presenters don’t know it or just don’t rate it. That’s the closest thing to drama here: not a flame war, but a subtle "why wasn’t our favorite invited to the party?" moment. The vibe is peak tech community: one person wants a video, one wants prettier graphs, and one wants a crossover episode.