Thank you for your support.
I am measuring MHz of my profile by "Manual Profile Layout".(no "Use Audio file input")
The result was very unstable (error seems -20% to +20%)
My understanding is that "Manual Profile" is calculated under static so it might be static.
Why it is not very unstable?
The Manual Profiling tool manually pumps each sublayout in a design with tuning commands without reliance on real-time audio interrupts.
It will be useful for me to know a bit more about your setup. What version of Audio Weaver are you using? What is your target platform? When you refer to "error," what are these values measuring and how are you determining them?
Thank you for your reply.
I attached some example (which is a small project (awd), then "error" which I mentioned was about 10%; 280MHz<->254MHz).
However I found that when I used another PC which was installed bigger memory (ie 128GB), the result of the Manual Profile Layout seemed stable. The attached example was with smaller memory PC (ie 16GB). I think Manual Profile depends on memory size of PC, does it?
When profiling an AWD in Native mode, it is common to observe variation in the MHz consumed. Core affinity and task priority, memory and memory bandwidth, even window size and ordering can affect the Native mode profiling results. You can tinker with these configurations and see how they change the profiling results (as you've done with comparing different PCs), but in general PCs aren't good real-time system for deriving meaningful profiling results.
Thank you for your explanation, I understood it well.
so,, To measure(pump) Manual Profile also needs to run the design not only "static calculation of the code", dose it?
Yes that is correct.
Sorry for my frequent question.
Can I refer the "MHz" of "Manual Profile" or/and "Profile Running Layout" which was retrieved in Native mode to estimate the actual CPU load of the target device?
Now we are considering which device is suitable for our system regarding CPU clock.
There is not an easy conversion between the Native Mode CPU load and the load on a different target processor. The best way to determine this resource consumption is to run the design on an embedded processor which has AWE Core integrated into it. For help and consultation on selecting an appropriate processor and comparing the load on different targets, please contact email@example.com.