I did that. The times reported here are are times for 1000 runs divided by 1000 !
On Thursday, June 26, 2014 12:45:21 AM UTC+2, Jason Merrill wrote: > > I think that sub-ms times are not really reliable. If you're interested in > accurate benchmarks, you might want to consider either running the code > many times in a loop (but make sure the optimizer doesn't eliminate the > actual calculation!), or better, using a larger problem instance.
