On Tuesday 11 April 2006 12:58, Loren Merritt wrote: > For whole codec optimizations, I get more reproducible results from > time(1) than from -benchmark. -benchmark seems to include time spent sleeping in the Sys value.