interpretation of profiler results

[Intel_C_Intel]

I do not understand some of the profiler results. The following are two lines of results from a run of my application through the profiler: (The data, from left to right is Func Time, %, Func+Child Time, %, Hit Count, and Function)

       Func          Func+Child           Hit   
        Time   %         Time      %      Count  Function   
---------------------------------------------------------   
   36945.772  17.2    36945.772  17.2 53983416 _exponent_d (expon_d.obj)   
   31007.156  14.4    67952.928  31.6 27400999 _DIVIDE_DP@8 (numrlib.obj)   

I know that the only routine in my program that calls _exponent_d is _DIVIDE_DP. The profiler states that the ratio of time in spent in the former to time spent in the latter is about 1.19. Based on this I would expect there to be the same ratio of hits in the former to hits in the latter. Instead I see almost twice as many hits in _exponent_d as hits in _DIVIDE_DP. Why? Are the "hits" not the result of a sample taken at equal time intervals? Which result, function time or hits, should I use to determine where my program is spending its time?

Thanks for you help.

Norm Clerman
Opcon Associates, Inc.

[Steven_L_Intel1]

I'm not all that familiar with the profiler, but I note in the documentation that "The Hit Count column reports the number of times that the function was called. " The documentation does say that with Function Timing, you get both time spent in the routine and number of times the routine was called.

Steve

[Intel_C_Intel]

That explains it. Thanks for your help.

Norm