I'm in the process of writing a logging library and I noticed that each call to print out ptime took about ~50us.  Even just printing out the duration time_of_day cost ~20us.

I'm using boost1.43 with g++ (GCC) 4.4.3
g++  -ftemplate-depth-128 -O3 -finline-functions -Wno-inline -Wall -pthread  -DNDEBUG

I was able to do the same thing for <1us not using stream ops (see code below).

Does anyone care to optimize this stuff?  I'm considering specializing my own boost::date_time::ostream_time_duration_formatter to override to slow impl.

template<typename int_t>
void printDecimal(int_t d,int places,char* out){
        int_t rest=d;
        for(int i=0;i<places;i++){
                out[places-i-1]= '0'+(rest%10);
                rest/=10;
        }
}

void printDuration(const boost::posix_time::time_duration& d,char* out){
        int64_t ticks=d.ticks();
        int64_t us=ticks%1000000;
        int64_t r1=ticks/1000000;
        int64_t  s=r1%60;
        int64_t r2=r1/60;
        int64_t  m=r2%60;
        int64_t  h=r2/60;
        printDecimal(h,2,out);
        out[2]=':';
        printDecimal(m,2,out+3);
        out[5]=':';
        printDecimal(s,2,out+6);
        out[8]='.';
        printDecimal(us,6,out+9);
}

..

                char buf[32];
                printDuration(pt.time_of_day(),buf);
                std::cout<<  buf <<" -\n"; 
                // <1us

                //std::cout<<  pt.time_of_day() <<" -\n"; // ~15us

Chris