>>>>Vlad
>>>
>>>All I did was to set up two FOR...ENDFOR loops, one that continually ORed the same numbers, and one that added them. By using the same literal values, and having no other code, I felt that it was the truest indicator of actual performance I could find. Any ideas on this?
>>>
>>>George
>>
>>I also did this. After that I switched the FORs. I got totally different results. So, I inserted another empty FOR before the other two FORs. Again, different results. I am a little confused and I don't know yet how to interpret these results. :(
>>
>>I noticed similar behaviors before, but in the other cases I found some explanations. Not this time.
>>
>>Vlad
>
>I've tried to simulate what you've done, and with the exception of the first test, there was little difference in the results, regardless of when the operations occurred. Using a 1000 iterations: addition consistently returned 0.003 seconds; OR was consistently either 0.004 and 0.005, with the exception of the first pass. The first pass was 0.037! I'm at a loss to explain this.
>
>George

I tried a million iterations, and the times came out as about 6.something seconds. The bitrshift was consistently a tenth of a second slower than multiply. (different operations I know)