Reply by Jerry Avins March 4, 20062006-03-04
Viswanath wrote:
> > Hi John, > > Thanks for your reply. > > You are saying to use the oscilloscope, but the problem is i don't have > oscilloscope with me.I have the DSK from which i have to do something. > > Is there any method such as clock() etc by which we can solve this problem > in the software itself. > > Thanks & Best regards, > Viswanath..
Extend John's method enough do a stop watch will do. Calling a loop from within a loop multiplies their execution times. With three nested loops, you ought to be able to reach seconds. If need be, slow the clock by a known factor. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by Viswanath March 4, 20062006-03-04
>
Hi John, Thanks for your reply. You are saying to use the oscilloscope, but the problem is i don't have oscilloscope with me.I have the DSK from which i have to do something. Is there any method such as clock() etc by which we can solve this problem in the software itself. Thanks & Best regards, Viswanath..
>Viswanath wrote: >> Hi all, >> >> I am doing my project on TMS320C5510.Upto now i am using profiling
method
>> to caluculate MIPS but sometimes it is not giving the correct reults
and
>> the results are differring.Is there any other alternative to
caluculate
>> MIPS. >> >> pls help me on this.. >> >> Thanks & Best regards, >> Viswanath.. > >Anytime I'm going to publish some execution times, I don't trust the >tools and go back to simple methods from before the days of profilers. >Here are a couple of suggestions: > >1. Profile a code section with known execution time, for example a RPT >#100 loop with a NOP in it (100 cycles). > >2. Blink an LED on every repetition of a code section with known >execution time and measure the blink interval on an oscilloscope. You >need to check your clock input and PLL settings. > >John > >
Reply by john March 3, 20062006-03-03
Viswanath wrote:
> Hi all, > > I am doing my project on TMS320C5510.Upto now i am using profiling method > to caluculate MIPS but sometimes it is not giving the correct reults and > the results are differring.Is there any other alternative to caluculate > MIPS. > > pls help me on this.. > > Thanks & Best regards, > Viswanath..
Anytime I'm going to publish some execution times, I don't trust the tools and go back to simple methods from before the days of profilers. Here are a couple of suggestions: 1. Profile a code section with known execution time, for example a RPT #100 loop with a NOP in it (100 cycles). 2. Blink an LED on every repetition of a code section with known execution time and measure the blink interval on an oscilloscope. You need to check your clock input and PLL settings. John
Reply by Viswanath March 3, 20062006-03-03


Hi all,

I am doing my project on TMS320C5510.Upto now i am using profiling method
to caluculate MIPS but sometimes it is not giving the correct reults and
the results are differring.Is there any other alternative to caluculate
MIPS.

pls help me on this..

Thanks & Best regards,
Viswanath..