Daniel, I asked you on Twitter a while ago if you thought a 32 x 18 (576 pixel) matrix would run fine off the AT32u4-based Yún and you said I might want to look into OctoWS2811; I didn’t quite understand why at the time but doing further reading I now realise the time it takes to push the pixel array down the whole strip is significant. Roughly how long is that per LED? Is this in the docs somewhere?
https://code.google.com/p/fastspi/wiki/ChipsetOverview has the data rates. At 800khz, it’s 1.25µs per bit, 24 bits per pixel, or 30µs per pixel. A 1000 pixel string will take 30ms.
Ah, I’d seen that page before but forgotten about it, thanks. So does that mean leds.show() will block for 30ms in the 1000 LED case?
Cool. Did the “grouping/blocking of multiple streams of output in parallel” that you mention on that page make it into FastSPI2 yet? If I split my large panel into two sections and ran them off different pins would that get me a higher frame rate, all things being equal?
i have the beginnings of it on a future tree - that project got put on hold trying to get 2.0 out the door, which in turn keeps getting stalled by life (personal projects, work, and most recently dealing with a death). Right now - running them on two different pins won’t get you a higher frame rate, because the software still has to write the outputs linearly - but I definitely want to get parallel output going!