On Wed, Sep 27, 2017 at 6:23 PM, Thorsten Ottosen <tottosen@dezide.com> wrote:
Benedek, could you please make a test of the performance of serialization for both devector/batch_deque vs boost::vector/boost::deque (release mode, full speed optimization), perhaps using the same measuring technique as employed by Joaquin. And then post results and code so people can run it on their favorite system. You should use types char, int, and something bigger, e.g. string or array<int,32>.
Hi, Please find the benchmark code and results attached. It compares serialization of devector/batch_deque/std::vector/std::deque. (I couldn't find a ready serializer for Boost.Container). The comparison is not exactly fair, because it seems boost/serialization/vector.hpp also adds version information. Measurement was done as above. Brief preview of the full set of attached results: ------------------------------------------------------------------------------------------------------------------------- Benchmark Time CPU Iterations ------------------------------------------------------------------------------------------------------------------------- boost::double_ended::devector<char>/32768 94197 ns 94039 ns 7430 std::vector<char>/32768 94351 ns 94235 ns 7457 boost::double_ended::batch_deque<std::array<char, 32>>>/32768 2975327 ns 2970627 ns 234 std::deque<std::array<char, 32>>>/32768 5237764 ns 5228441 ns 133 Benedek