22 Sep
2011
22 Sep
'11
9:36 a.m.
On 22/09/2011 11:25, Sebastian Redl wrote:
On 21.09.2011 20:20, Mathias Gaunard wrote:
As part of the Boost.SIMD effort, another library was developed, Boost.Dispatch, which relies heavily on overload resolution.
Back to the benchmark, making the number of functions vary show an apparent exponential increase in compile time. It's reasonably fast for 100 functions, but not at all for 1,000.
Overload resolution is supposed to be linear in the number of overloads. If you can show a graph that shows super-linear increase in compile time, you should file a bug with the respective compilers.
Will do. a quick Excel regression show a 0.98 fitness to a+b^X ...