← Feature Requests

Build for PiStomp / other pi5 based devices

K. S.10/12/2026, 3:22:49 AM

Hi I came across your plugin on the Mod devices forum.

Is there any chance you would be willing to make a build of your plugin for other devices that use the mod ui?

I use a PiStomp (https://www.treefallsound.com/) and it runs .lv2 plugins compiled for arm64, with the same interface as mod devices.

Most of our plugins are uploaded to patchstorage (https://patchstorage.com/platform/lv2-plugins/) and we can directly download them from there.

Thanks for your help

would love to be able to use your awesome plugin on the pistomp.

Ken.

gerald mwangi10/12/2026, 8:53:17 AM

Hi. Thanks for reaching out. I would indeed be willing but there are some constraints:

I dont own the pistomp, not even a pi5 yet.

I dont have the means yet to buy the hardware since I am working full time on the plugin and dont have revenue yet.

I would need financial support (See the donations section at https://github.com/geraldmwangi/GuitarMidi-LV2#donations).

So if you could donate the device I would put in the effort to port the plugin.

My work on the official port to the mod devices is very positive for the mod duoX. Im sure it will work on the pi5 since pi5 has a far stronger cpu

!!!However I cannot guarantee that!!!

gerald mwangi10/12/2026, 9:01:57 AM

Have you tried the plugin out on linux?

K. S.10/15/2026, 11:02:05 PM

I have it built and runnnig on the piStomp. Have not had much chance to test yet, will hopfully be able to report more on this over the weekend.

gerald mwangi10/16/2026, 8:00:42 AM

Thats great 😀. Looking forward to hearing about the test. Thanks for your work

K. S.10/17/2026, 9:49:11 AM

I've had some success.

The initial build was using 99% cpu on one thread, and 20%-50% on the other three. Turns out that the polling for new audio was the culprit.
Initially I added a wait (100-200us) in the polling loop, and this worked pretty well.
Using some AI help, I made some modifications to modelinferencer.hpp and modelinferencer.cpp to make it event driven inferencing, when a new packet of audio is ready, a semaphore is triggered.

This reduced load to about 20-25% in one CPU and 17-20% on the other 3.

Also updated the modGui to look more like your desktop version.

My fork is here.

https://github.com/tothoseofus/GuitarMidi-LV2

Have a look, happy to contribute back.

Edit: All of the changes should be compatible with your master, except maybe som of the changes in the cmake files. I was having trouble with the Zita-Resampler.

gerald mwangi10/17/2026, 5:02:43 PM

Hi. Thanks ill check your repo in monday.

How is the tracking?

Log in to reply.