Midi modules

[JoPo]

As it's difficult to check only with the dsp meter, I'd like to know if a module like midi filter takes a lot of dsp power.

Maybe, people who developed some midi device can give me an answer..?

Thanks a lot !

[garyb]

no it does not use much dsp at all.

[sharc]

Most MIDI modules use very little DSP in terms of cycles. You can fit more than 350x MIDI Filter modules on a single DSP.

Usage of DSP memory is more of an issue with some MIDI modules - Not the MIDI Filter though.

For example the module used to handle MIDI CC assignments in devices will use up all available DSP memory quicker than most DSP modules.

[jksuperstar]

Roy did a great job of assessing module use of DSP for modular, I thought he once had more info, but seems less now. Still, look at the Modular section of cwmodular.org, and check both MIDI and Seq areas. Most modules use 0% if async operation is used.

[JoPo]

Thanks for the info !

Midi filter is light, but modules which generate midi lfo like there is in Wolf midibox are much more dsp hungry, no ?

-->Shark--> yes ! I noticed that : if I send a lot of midi control data to Scope devices, Scope becomes very very slow ! Besides, that's why I have a lot of interest about midi filter : I say to my smart self that with midi filter to blok midi data Scope devices don't need, those devices will take life much easier !

Am I right ? Am I as smart as I imagine ?

[sharc]

Midi filter is light, but modules which generate midi lfo like there is in Wolf midibox are much more dsp hungry, no ?

In Scope there's a DSP atom which is used to handle MIDI CCs in devices. This atom uses quite a lot of (much more than average) DSP memory.

Here on my 3xDSP card, I can only load 15 of those before DSP is maxed out.

Basically, any regular Scope device which has MIDI IO connections AND gives access to the MIDI Ctrl Assignment window via right-clicking a control is using this atom.

It's worth pointing out here that modules within a modular patch and insert FX within Mixers etc. are usually using the host device for CC assignments so won't be adding any extra load on this front.

-->Shark--> yes ! I noticed that : if I send a lot of midi control data to Scope devices, Scope becomes very very slow ! Besides, that's why I have a lot of interest about midi filter : I say to my smart self that with midi filter to blok midi data Scope devices don't need, those devices will take life much easier !

This issue is a bit more complex as it involves more variables. It usually comes down to what devices you have loaded and what group of parameters you're controlling. Automating some parameters will involve more system processes than others.

It's probably not a bad idea to filter out unnecessary MIDI data. I'd recommend trying something like MIDI OX (that MIDI data won't then be handled by Scope at all) for this task and see how it compares with the Scope MIDI filter.

[JoPo]

It's worth pointing out here that modules within a modular patch and insert FX within Mixers etc. are usually using the host device for CC assignments so won't be adding any extra load on this front.

Very interesting to know !

Yes sure for midi ox but isn't midi ox, in turn, going to eat all my PC ressource ? - But as far as I know midi data computing is not very CPU consumer for a PC, is it ?

Anyway, if you advice me to stop midi data before it goes in Scope, it's worth a try... Thanks a lot !

[sharc]

It's worth pointing out here that modules within a modular patch and insert FX within Mixers etc. are usually using the host device for CC assignments so won't be adding any extra load on this front.

For the sake of avoiding any confusion, 'host' in this context means 'parent' Scope device as opposed to the OS.

In my experience of using MIDI OX it seems to use very little in the way of resources. I'd imagine the likes of CopperLan would be a possible alternative, but have to admit I've never done any intense stress testing on either.

The point here though is that the Scope MIDI input isn't the issue as much, but the Scope Sequencer MIDI Input -> MIDI filter stage might have more of an impact on performance than filtering the MIDI data beforehand. If I knew this for certain one way or the other I would say so. It's just a suggestion

[JoPo]

Excellent, Shark, all those info, new for me.
Now, I need to make some tests.

Thanks a lot !

[JoPo]

Since now, if I add 'send to' Scope in a Cubase vst instrument midi track to control the sound of this vsti in scope and have the controler track editing in the same midi track, I place the 'transformer' Cubase midi fx on the send to keep only midi data really needed for the Scope sound control.

Like this, it's upstream, as Shark recommended : I filter midi data before it reach Scope and I can notice that Scope is very happy to receive as less midi data as possible.

BUT (info for Cubase 8 users) : when you add a logical 'or' in the transformer, you must save the preset, close it and reload it again with the preset to have it working ! A Transformer bug which I took time to find !

transform.jpg (65.16 KiB) Viewed 2045 times

In this example (sorry in french) , I let go to Scope only controler 12 & 13, no note and no pitchbend data.

Thanks for your attention.