Esp32p4 copy tiles to framebuffer speed improvement

[themindfactory]

I have an esp32 p4 that's connected to a 1280x800 display running 2 Lane mipi at about 750 MB per second per Lane. Of course the frame buffer is in psram 2 MB and I have all sorts of tiles 100x100px that I would like to write to the screen as fast as possible and they all reside also in psram I am not getting throughput like I would like. Are there any tricks to know? I have tried the PPA I've tried straight mem copies. I found PPA slower than the mem copies. Not sure. Is there anything I should know about psram alignment cache memory alignment? What is the faster way to do this? Btw it's 565 format data.

Richard

[MicroController]

What's your speed set to?

[themindfactory]

Speed 52fps on the mipi and 360mhz uC clock 200mhz psram