2 Feb
2009
2 Feb
'09
7:17 p.m.
On Mon, Feb 2, 2009 at 14:02, Edouard A. <edouard@fausse.info> wrote:
I haven't benchmarked against a runtime option, I did a template parameter as it felt more natural to me to make this a compile time parameter rather than a runtime parameter.
It seems fundamentally runtime to me, since different machines or just different runs will want different levels of concurrency. The overhead ought to just be a few compares and arithmetic operations, which would be swamped by the effort involved in the sorting. I think normally I'd want to just use parallel_sort(b, e, thread::hardware_concurrency())