🚀 go-pugleaf

In article <1992Sep30.012142.25740@nuscc.nus.sg> eletanjm@nuscc.nus.sg (TAN JIN MENG) writes:
>schmke@cco.caltech.edu (Kevin Todd Schmidt) writes:
>: 47@unpcs1.cs.unp.ac.za (menglund) writes:
>: 
>: >I am wanting to know how to determine a machines clock frequency dynamically
>: >within a program. (eg: 8MHz, 16MHz ect).
>: 
>: The way I do it is to measure the time it takes to execute a big block
>: of NOP's.  You may need to loop around the block a few times to get
>: an accurate measurement.  Then, based on the number of NOP's executed
>: you can calculate the clock speed since you know the number of clocks
>: needed for a NOP to execute.
>: 
>: If you have any questions let me know.
>: 
>: Kevin
>: -- 
>: Jet Propulsion Laboratory | schmke@cco.caltech.edu
>: 4800 Oak Grove Dr.        | schmidt@spc5.jpl.nasa.gov
>: M/S 525-3684              |
>: Pasadena, CA  91109       |
>
>I would think that with this method, you'd need to account for
>external/internal cache also.
>
>One way, I suppose is to discard the first few iterations of the loop.
>
>Hmm?

This assumes that the "big block of NOP's" is smaller than the size
of the on-chip cache or even that matter the second level cache.  
Otherwise it won't matter if use or don't use the first iterations. 



Rod Skinner				I speak for myself only.
Intel Corp
3065 Bowers Avenue   MS NW1-20            PH (408) 765-4474
Santa Clara, CA 95052                     FAX (408) 765-4920