Hi All, I am trying to use feedback directed optimization (fprofile-use) in gcc. I just wanted to know what the general code shipping procedure is here? Can we just run the instrumented binary (built using fprofile-generate) once on training input and check in the generated gcda files to the repo? So, anyone who has to compile again will checkout the gcda files along with the code. And compiling the code with these generated gcda files will produce the optimized binary (as long as the new code path is same as the original instrumented code path) ? Are the gcda files portable this way ? Thanking You, Anant Sah
