Info table profiling is a new profiling method in GHC that enables attribution of heap allocations to specific source locations in a Haskell program. Notably, this profiling method does not require that the profiled program is compiled using GHC’s profiling mode. In combination with tools such as eventlog2html and ghc-debug, it can help us answer very precise and low-level questions about a program’s memory behavior.

While info table profiling is extremely useful for analyzing and debugging the memory behavior of Haskell programs, it can result in significantly slower compile times and larger executables. In this talk, I will present some recent improvements to GHC’s info table profiling capabilities that address these issues, and briefly discuss the effect these modifications can have on common profiling workflows with eventlog2html and ghc-debug. This talk is aimed at advanced Haskell developers who are interested in how these profiling methods and improvements might apply to their Haskell programs.