Hello,
first of all thanks for all the efforts that you are putting into the development of LVGL!!
I’ve been playing around with LVGL on a STM32U5G9J-DK2 demo board with NEMA acceleration. I have configured the board mixing information taken from the official LVGL port for the Riverdi board (GitHub - lvgl/lv_port_riverdi_stm32u5) and from a project created in with TouchGfx
I’ve been trying to find some results for the LVGL benchmarks on the Riverdi board or on another U5 board with and without NEMA enabled, to understand if I’m on the right track in configuring my project, but I could not find anything…
Could you please point me to a page with some benchmarks or maybe share your results?
Also a couple of additional questions:
- Since there is plenty of internal RAM on the chip I’ve configured a full double buffer in direct mode using
lv_st_ltdc_create_directand enabled the LTDC driver built-in into LVGL. Is this solution offering the best performance for this chip? Or would using a partial buffer be better also in this scenario? - According to LVGL docs NemaGFX Acceleration — LVGL documentation
32 and 16 bit LV_COLOR_DEPTH is supported. I’m using 24 bit and everything seems to be working correctly. Is this wrong and if yes why 24 bit is not supported?
As a reference here are the results that I have at the moment. I’ve set the target FPS to 60 to see how much room there is on the board (even if it makes not that much visual difference with respect to 30 FPS)
Unfortunately I’m still a beginner with both LVGL and ST hardware so I don’t really know how to rate these numbers 
