Shalom<br><br>I&#39;ve updated this page --<br><a href="http://webEbenezer.net/comparison.html">http://webEbenezer.net/comparison.html</a><br><br>with results from Boost 1.52.    <br><br>A few notes about the testing:<br><br>
I decided to use -O2 on the Linux/g++ tests.  In previous<br>tests I used -O3.<br><br>I hadn&#39;t updated the Windows test results since the  <br>Boost 1.48 time-frame.  Around Boost 1.51 I had to <br>start building the serialization library in Boost <br>
with runtime-link=static in order to get the Windows<br>tests to build.  I haven&#39;t understood why that became<br>necessary, but the results for the serialization library <br>in Boost seem to have been adversely affected by this.  <br>
On Windows, I now report that tests built with the <br>serialization library in Boost are over 1.8 times slower <br>than parallel Ebenezer tests.  Previously I reported the <br>tests built with the serialization library in Boost were <br>
between 1.7 and 1.8 times slower than the Ebenezer tests.<br><br>--<br>Brian<br>Ebenezer Enterprises -- making programming fun again.<br><a href="http://webEbenezer.net">http://webEbenezer.net</a><br><br>