How to use the Profile Clock for C55x.

How to use the Profile Clock for C55x.

I want to see the right usage of the Profile Clock.

I am using TMDX5535EZDSP in CCSv4.

CCS menu to enable the Profile Clock is "Target - Clock - Enable".
CCS menu to reset the Profile Clock is "Target - Clock - Reset".

I used the Profile Clock in two case for the following assembler cord.

       mov hi(ac0),*ar1 || mov ac0,*ar3
       mpym *ar0+,uns(*ar3+),ac0
       macm uns(*ar0+),*ar1+,ac0
       macm *ar0+,uns(*ar3+),ac0
       macm uns(*ar0+),*ar1+,ac0
       macm *ar0+,uns(*(ar3+t0)),ac0
       macm uns(*ar0+),*(ar1+t0),ac0
       macm *ar0+,uns(*ar3+),ac0
       macm uns(*ar0+),*ar1+,ac0
       macm *ar0+,uns(*(ar3-t0)),ac0
       macm uns(*ar0),*ar1,ac0

       amov #PreEQCoef1,ar0
       asub #5,ar1
       macm *(ar0+t0),*ar1+,ac0>>#16
       macm *(ar0+t0),*ar1+,ac0
       macm *(ar0+t0),*(ar1+t0),ac0
       macm *(ar0+t0),*ar1+,ac0
       macm *(ar0+t0),*(ar1-t0),ac0
       sfts ac0,#2       || mov #4,t1      
       sat ac0

       mov hi(ac0),*ar1 || mov ac0,*ar3

The first case, I put a breakpoint in the first line and the last next line each and ran the program.
When stop at the breakpoint of the first line, I reset the count of the Profile Clock.
When stop at the breakpoint of the last next line, I watched a result, it is 24.

The second case, I run the program by single-step.
When stop at the first line, I reset the count of the Profile Clock.
When stop at the last next line, I watched a result, it is 21.

Why is it that results were different?

Best regards,
Daisuke