Just want to get others’ takes on MNodes performance. Am using Danny’s 3 OSC synth in a Studio One project on a Mac and it seems that MNodes seems to be somewhat heavier CPU-wise than other VST tools I am using. Here you can see the performance of Danny’s patch in MNodes relative to IK Multimedia’s Hammond B-3X and Arturia’s Pigments. All 3 were playing the same midi clip of just mostly triad chords. I’m not sure what I’m saying here but I really want MNodes to be super successful and I just felt like it was important to point out current CPU usage (probably not new news to many here on the forum). Perhaps more optimization will take place in the future but I’m just a tiny bit concerned about it. Thoughts from others? Maybe its just my problem. 
Hi Jeff,
I have experienced in a previous version to a newer on that MNode has improved a lot so i do wonder if can be optimized even more. Lets hope so.
But i have to say, i did not use the polyphony module which could deliver a better performance. there are some reason’s why i didn’t use that.
In general i find the MNode performance good for a lot of things but i must say, i am also some type of person that always love to go bigger and see if i hit some limits.
And with this, me experimenting there will be this type of issue. So should i went of a 3 OSC, perhaps not. but i wanted to see how much CPU usage it cost’s and what i could do to reduce it.
So i hope, that’s my goal.. to improve my project also. with or without MNode’s performance in the future. just keep in mind that my project is also just an experiment from a person who is investigate what MNode can or can’t do at this moment.
PS: For me the Cpu usage is also high. that high that i even would not run pigment or another instruments. I would likely bounce the audio of that instrument and disable it 
Thanks for the good input. I think any serious instrument built with MNodes will be at least as complex as your synth patch if not more so. Using Reaktor as an indicator, it is not at all unusual for good quality instruments to be quite complex under the hood as one can see by drilling down into the internals of those patches. I would guess that MNodes patches will be no different on the complexity spectrum. Hopefully MNodes will be a great tool for those wanting to build deep and flexible instruments and effects without taxing our systems to the point of people looking elsewhere for lower CPU overhead. That said, I’m still a huge fan of MNodes and I don’t see that changing.
i wonder how yours performs without any filmstrip ive found a very simple project with filstrip use much more cpu than my very complex harmonics eq where i just use very basic controls
Thanks for the report. I’m looking into it thoroughly.
Danny’s project is truly massive, and of course, with any project of this size there is always some responsibility on the developer’s side (in this case, Danny) to optimize it. That said, MNodes should definitely be able to handle projects of this scale much more efficiently.
After reviewing and optimizing things, I managed to reduce CPU usage on Danny’s patch from around 50% down to 30%, which is a pretty significant improvement. I’ve also dramatically reduced the loading time, as I found an issue related to Filmstrip handling.
I’ll keep refining the engine and optimizing it as much as possible. I’ll do my best to have the new version ready as soon as I can.
In general, I suspect that something like MNodes will always require more CPU than a plugin compiled from C++ written and optimized for a specific task. What you get in return is ultimate versatility and customization.
I have no issue’s in performance with the filmstrip node. As a resource i have 4 filmstrips for knobs and every source is used 25 times, so 100 filmstrip widgets. At the moment when my filmstrips where more around 10mb each the loading time of the project with the latest update went from 2 seconds to if i remember well 35 seconds.
While using this project while i am testing things out my is the only DSP usage around 30%-40% while the real CPU usage is scaling between 0.5% and 9% depending on the oscillator/instrument types.
Using any filmstrip knob which are now optimized to 1.33mb and 128 frames are running smoothly and have no CPU impact while using them. the loading time of the project is reasonable reduced that the loading time went down to 15 seconds.
My system specs are i5-12600k, 32gb ram, Nvidia 4070ti and i only have SSD’s.
Great analysis Danny. Thanks for sharing.
well yeah but that is you here things are different
Of course, but since you replied to me while you where asking about my performance i thought to share my system spec’s for you to see if anything matters. Some people are running a onboard GPU for there daily work and share that memory/power with the CPU.
There are enough situations where a motherboard has an update and needs also some new intel drivers for example.
Try this one, i optimized a lot and using the poly module so it scales better on the actual voices played.
Good to know how this is changed for you, makes the project more viable to be used.
Thanks for the feedback! 
Incredible, Danny!!
I’ve been testing your patch with the current version (1.0.3) against the new one I’ll be releasing soon (1.0.4), and the new version actually reduces CPU usage even further! Hopefully I’ll be able to get it ready for release today.
Thanks Guys!