Re: [v3][PATCH] Framework for performance benchmarking of functions

[Richard Henderson <rth at twiddle dot net>]

On 2013-02-20 21:22, Siddhesh Poyarekar wrote:
> I wonder if it makes sense to calibrate the benchmarks based on an
> estimate of how long a clock_gettime takes, perhaps by doing
> consecutive clock_gettime calls (multiple times) and seeing results.

If you're thinking in that direction, have a look at gmp's tune/speed 
program, and the rather large lengths they go through to eliminate 
overheads.


r~