This takes much of the guesswork and trial-and-error out of debugging and means that bugs that might otherwise have taken weeks to fix can now be fixed in minutes. UndoDB works on Linux programs running on Intel/AMD processors and (soon) on ARM.
The Undo reverse-execution engine is compatible with gdb 7, and plugs seamlessly into gdb's 'process record' feature. Developers can use UndoDB wherever they use gdb (e.g. at the command-line, within emacs, ddd, or Eclipse). The result is similar functionality to regular gdb, but with recording performance many thousands of times better: UndoDB's slowdown is typically less than 2x, compared to gdb's recording slowdown of more than 50,000x. UndoDB is used daily on some of the world's most complex software: from scientific computing (NASA, LLNL), to enterprise (including some of the world's largest banks) and design automation (including the world's largest EDA vendors). A study by Cambridge University in 2012 found our customers report that UndoDB improves debugging productivity by an average of 26%.