StefanPetrick
Well-known member
Hi everybody!
I looks like my little LED animation projects turns out to become a full blown modular animation synthesizer. For consumer size setups with 1k LEDs one Teensy 4 will be more than enough for +-10 animation layers and complex animations.
But I have also have some people reaching out already with 10k - 40k LEDs setups. They adapted my code and have ESP32S3s running it on 2 cores + FPU now. Framerates close to one T3.6, sometimes even more leveraging some tricks.
I wonder: for a top notch (high framerate) setup - wouldn't it be the best to have multiple T4s receiving parameters from one microcontroller - and rendering the data required for the whole LED matrix - all on individual nodes fed by one single master controller which handles the distribution of the controlling parameter data so that the slave units each generate RGB data for their individual sector of the physical LEDs based on the current animation layer parameters?!
Long story short, my question is: what`s the fastest protocol I could use? I wish to have a T4 speak to some other T4s one way only, ideally simultaneously, in order to distribute all required parameters and timestamp data to multiple T4s. I want to enable physical and logical parallelizing of the rendering process itself - for high fps results.
What should I look into? SPI? I2C? CAN? Something else? What would be the prefered weapon of my choice for my specific usecase?
Requirement: I need to send +-100 times / second a data packet containing +-2kB of of control data to multiple receivers ASAP. Every µs not sped on waiting for - or receiving of - data can be spend on rendering more RGB data frames instead.
Latest experiment:
I looks like my little LED animation projects turns out to become a full blown modular animation synthesizer. For consumer size setups with 1k LEDs one Teensy 4 will be more than enough for +-10 animation layers and complex animations.
But I have also have some people reaching out already with 10k - 40k LEDs setups. They adapted my code and have ESP32S3s running it on 2 cores + FPU now. Framerates close to one T3.6, sometimes even more leveraging some tricks.
I wonder: for a top notch (high framerate) setup - wouldn't it be the best to have multiple T4s receiving parameters from one microcontroller - and rendering the data required for the whole LED matrix - all on individual nodes fed by one single master controller which handles the distribution of the controlling parameter data so that the slave units each generate RGB data for their individual sector of the physical LEDs based on the current animation layer parameters?!
Long story short, my question is: what`s the fastest protocol I could use? I wish to have a T4 speak to some other T4s one way only, ideally simultaneously, in order to distribute all required parameters and timestamp data to multiple T4s. I want to enable physical and logical parallelizing of the rendering process itself - for high fps results.
What should I look into? SPI? I2C? CAN? Something else? What would be the prefered weapon of my choice for my specific usecase?
Requirement: I need to send +-100 times / second a data packet containing +-2kB of of control data to multiple receivers ASAP. Every µs not sped on waiting for - or receiving of - data can be spend on rendering more RGB data frames instead.
Latest experiment:
Last edited: