Ffl ffu

[ASF]

Mellis is right - for the most part, I just use COP on anything other than asynchronous data, as CPS is a whole lot more CPU intensive, but if it's possible that your instruction could be interrupted halfway through, and that the half-finished data could be used somewhere else, and if that would cause a problem, then you're best to use CPS.

 

[Pete.S.]

The fastest and most elegant solution when you have large arrays is to NOT shift any values at all, but to use a ring buffer.

It's very easy really. It's just an array. All you need to keep track of is where the first (oldest) value and the last (newest) value is. No moving of any data required. As you come to the end of the array you start at the beginning.

It's perfect for keeping track of things moving on conveyers or keeping a history of things.

https://en.wikipedia.org/wiki/Circular_buffer

So to put a new value in

Newest=Newest+1
If Newest>=100 then Newest=0
Array[Newest]=value

Ta take the oldest value out

value=Array[Oldest]
Oldest=Oldest+1
If Oldest>=100 then Oldest=0

Look at the ten newest value

x=Newest-10
if x<0 then x=x+100
value=Array[x]

If you're lazy you just put 256 values in the array and use an unsigned byte to keep track. Then you can skip the check if you are outisde the array becuase the byte will take care of that by itself.

 

[daba]

Mellis is right - for the most part, I just use COP on anything other than asynchronous data, as CPS is a whole lot more CPU intensive, but if it's possible that your instruction could be interrupted halfway through, and that the half-finished data could be used somewhere else, and if that would cause a problem, then you're best to use CPS.

I disagree, CPS is no more CPU intensive than COP, they do exactly the same job in the same way. CPS just prevents "interrupts" from higher priority tasks while it is executing, so guarantees a valid "data-set". This disabling and re-enabling of interrupts adds a tad to the execution time, but I certainly would not call it "a whole lot more CPU intensive".

CPS is the equivalent of UID COP UIE

I use CPS by default, unless I can think of a good reason not to.....

 

[ASF]

I disagree, CPS is no more CPU intensive than COP, they do exactly the same job in the same way. CPS just prevents "interrupts" from higher priority tasks while it is executing, so guarantees a valid "data-set". This disabling and re-enabling of interrupts adds a tad to the execution time, but I certainly would not call it "a whole lot more CPU intensive".

The knowledge base disagrees...have a look at Technote 50235 (techconnect required), and the attached Word document. It's quite detailed and interesting, but the upshot of it all that I'm referring to is on page 2:

When compared to the COP instruction which only performs step 4, you can see there is a tremendous amount of overhead for the CPS instruction. Because of this it should only be used when dealing with I/O and Produce/Consume data.

There's also another technote (Technote 46251, Techconnect required) which details how to choose between COP and CPS - ultimately it ends up with "in this very specific situation, use CPS; otherwise use COP".

I mean in 99% of applications it's probably all academic with the processing power of modern controllers

 

[daba]

The knowledge base disagrees...have a look at Technote 50235 (techconnect required), and the attached Word document. It's quite detailed and interesting, but the upshot of it all that I'm referring to is on page 2:


There's also another technote (Technote 46251, Techconnect required) which details how to choose between COP and CPS - ultimately it ends up with "in this very specific situation, use CPS; otherwise use COP".

I mean in 99% of applications it's probably all academic with the processing power of modern controllers

I stand corrected....