Compiler-Assisted Floating-Point Error Analysis and Profiling with FPChecker
3 days ago
- #Scientific Computing
- #Compiler Tools
- #Floating-Point Analysis
- Half-day tutorial on June 22, 2026, from 9:00am to 1pm Europe/Berlin time, held in Room Hall X12 at CCH – Congress Center Hamburg.
- Introduces FPChecker framework and clang/LLVM tools to instrument C/C++ code for analyzing floating-point arithmetic errors, including rounding effects and error propagation.
- Covers evaluating dynamic range and precision requirements, tracking errors across code regions, and detecting infinities/NaNs in examples like linear solvers and finite-difference methods.
- Includes hands-on sessions with tools distributed via Conda and pre-configured AWS instances, with an advanced session for applying tools to real applications or libraries.