Strange CPU Profile

[James_McN]

I'm posting this here as it is more of a curiosity and I wonder if anyone else has seen this.

I've recently been doing some profiling of an embedded application running on a cRIO 9082 (dual core i7 running Pharlap).

In the graph the x axis is minutes and the total axis is 24hrs. For some reason we get a strangle rise and fall of CPU (don't worry the big spike is expected).

I'm not sure where to start or even whether to bother investigating further since it is all within tolerance! But if you have seen something similar I would love to know.

[MikeeB]

What does the memory usage look like? Does it track this by any chance?  Mike

[Simplicity_Jon]

Is the other CPU flat? Or does it exhibit the same sort of shape?

I have found with the dual/quad core controllers that individual core usage can jump about quite a bit especially when you have multiple loops running simultaneously. Overall the CPU usage is flat but LabVIEW just chops and changes which cores parts of the code gets executed on.

Jon

[swatts]

It's dreaming

[Simplicity_Jon]

of electric sheep!

[James_McN]

Memory usage has small variations but nothing that seems to track this:

Both CPUs are doing the same thing! In fact, both CPUs are identical which is suspicious! Although these are averaged over a minute, it still seems suprising they are so close. Will check my averaging code for any problems which may be causing some of these wierd effects!

[Pie566942.0]

Hi James,

I have no background with the 9082 BIOS, and I know you know what you're doing, so this may be way off.  Depending on how the CPUs are configured for RT, Intel might throttle them for stability and performance across temperature. I'd expect less gradual and more step-like transitions, such as the four saw tooth steps as the CPUs throttle and SMP reassigns. Can you correlate this to anything in the environment, specifically temperature?

-Steve K