TTD uses code instrumentation so typically has a lot higher overhead than rr on single-threaded code. On the other hand, it can handle shared memory and use mul

TTD uses code instrumentation so typically has a lot higher overhead than rr on single-threaded code. On the other hand, it can handle shared memory and use multicore, so for parallel programs on big machines TTD could be faster. TTD has some debugging features rr+gdb doesn't have, like some cool query-based debugging features. But it doesn't let you run application functions (e.g. prettyprinters) at debug time like rr does. Pernosco consumes rr recordings and blows TTD away IMHO :-).