CPU load question

I was wondering if any of you engineer-types could answer a question for me: In a modeling device such as the Axe-Fx, Helix, Kemper, etc., is the majority of CPU load taken up by active amp models/effects, or do the amp models/effects in their inactive states contribute as well?

Thanks!
 
This is really a DSP question. I doubt if the pipeline is not needing to do anything, it will not do anything. But most DSPs have a hard limit concerning how many operations they could do and the programmer(s) need to draw a line concerning max operations allowed.
 
For something like the Axe-FX usually once a module is loaded, it is using processor time. So even if you have bypassed modules, they will still be computing, but it won't necessarily have an audible effect. The reason being that even bypassed they may have internal state that needs updating, for example if you have a reverb module and then you bypass / mute it, and then unmute it, it would make sense that the reverb was still decaying during the time it was bypassed as opposed to frozen or just being dumped. It also makes it easier to allocate resources for all the modules that could potentially be active at once, you won't run into a situation where activating a particular combination of modules will make you short on CPU and start causing audio dropouts. Furthermore it allows you to do some optimizations in some cases. It's easier to just set some gain coefficients to 0 rather than restructuring your code and/or dealing with a bunch of branches (branching is pretty undesirable in DSP).
 
Kind of curious why this info is needed?

BTW as a comparison, on an idle Mac laptop, open up the activity monitor app and be astounded.
 
I was wondering if any of you engineer-types could answer a question for me: In a modeling device such as the Axe-Fx, Helix, Kemper, etc., is the majority of CPU load taken up by active amp models/effects, or do the amp models/effects in their inactive states contribute as well?

Thanks!
You can see easily that CPU increases when a block is added, either bypassed or active.
 
You can see easily that CPU increases when a block is added, either bypassed or active.

Yes, I guess I should have been more specific by what I meant by “inactive.” I was curious as to whether or not a block that hasn’t been loaded into a preset has any effect on the total CPU available.
 
Kind of curious why this info is needed?

BTW as a comparison, on an idle Mac laptop, open up the activity monitor app and be astounded.

Well, I been envisioning some kind of device (modeler, multi-fx processor, etc.) with the ability to customize what kind of effects or models are loaded into it. This way the user could maximize the device’s processing power for things that they wish to use. Let’s use the FM3 (so excited!) as an example. Some people are not liking the fact that you can’t have two amp blocks. If you could remove blocks from the FM3 that you don’t use, like maybe the Looper block or pitch block, and free up extra processing power, then perhaps an additional amp block could be added.

Of course, this would be contingent on whether or not a block that isn’t loaded into a preset has any bearing on what the DSP is capable of handling, hence my curiosity.
 
Yes, I guess I should have been more specific by what I meant by “inactive.” I was curious as to whether or not a block that hasn’t been loaded into a preset has any effect on the total CPU available.
Since you can see the cpu change when you add a block, it can be derived that not adding a block doesn’t use CPU.

Do you mean memory or CPU though?
 
Back
Top Bottom