This question evoked no response from the binutils list nor from the last known address of gprof's author. The data collection part of gprof is in GCC, so this is the logical next place to ask. In the info documentation for gprof, Section 6.1 (Statistical Sampling Error) states, "For N samples, the expected error is the square-root of N." However, the derivation or source of this result is not cited. I would be grateful for any information that would help to explain how this result was obtained or to locate the person who could explain it. The result does not appear in either of the papers associated with the earlier Berkeley gprof. Thank you, -- David Flater, National Institute of Standards and Technology, U.S.A.
