On 6/11/14, 2:58 PM, Vladimir Batov wrote:
1. The changes in spirit_new_test compared to the original spirit_int_test were additional conversion validity checks and the std::string... the same way boost::cnv::spirit works. That slowed spirit_new_test some 11% compared to spirit_int_test. Though raw conversion speed is important (spirit_int_test) I do not expect it deployed like that (with no checks) in a realistic setting.
2. boost::cnv::spirit implements the same process-flow as spirit_new_test and boost::cnv::spirit was consistently 6% slower compared to raw spirit conversion deployment with the same functionality (spirit_new_test).
Thank you, Joel, for sharing your performance testing framework.
My pleasure. I'm glad it helped. BTW, as I said before, you can use the low-level spirit int and real parsers without going through the compile-time consuming expression template facade. For example, here's how integers are parsed: http://tinyurl.com/mb5k9ql Notice: typedef extract_int<T, Radix, MinDigits, MaxDigits> extract; return extract::call(first, last, attr_); You only need to include: #include <boost/spirit/home/qi/numeric/numeric_utils.hpp> The same for unsigned ints and reals. It's all in: boost/spirit/home/qi/numeric/ unsigned ints: http://tinyurl.com/orl62mc floating pt: http://tinyurl.com/nu8vkhh Rip the code in the 'parse' functions for your own use. Then you don't need to wait 22 billion years for the compiler. I won't be surprised if you can still save a few cycles by going lower level, although the compiler should've optimized the extra scaffolding already. Regards, -- Joel de Guzman http://www.ciere.com http://boost-spirit.com http://www.cycfi.com/