On 14/10/10 7:30 PM, bearophile wrote:
Peter Alexander:
That way, you're actually seeing what move semantics will
really bring to the performance of a real program.
There are synthetic benchmarks, and various shades of more realistic
benchmarks. Both kinds have advantages and disadvantages, and both have their
place and may be useful. That's a synthetic benchmark.
Bye,
bearophile
Fair enough, although I find this benchmark pretty useless. All it says
is that move semantics can provide performance improvements in contrived
situations. You can show demonstrate just about anything in contrived
situations.