Hi,
I’m using LVGL on a system with an NXP i.MX RT1170 CM7 and a 480x800 display (natively in portrait mode). In our application we’d like to rotate frames and we currently do so by rendering the full buffers (LV_DISPLAY_RENDER_MODE_DIRECT) and rotating the full buffers using PXP.
I’m now investigating options for improving frame rate and I had the thought to use LV_DISPLAY_RENDER_MODE_PARTIAL (which means we can draw to a 1/8 size buffer in much faster memory) and rotate only the updated parts with PXP directly into the frame buffer. Unfortunately the output buffer locations (a subset of the full frame buffer) will no longer be aligned to 64 bytes and PXP truncates the results. Note: rotating only the updated areas works correctly with lv_draw_sw_rotate() except it’s too slow.
Do you have any suggestions to make this work, or other ways to improve frame rate?
Best,
Kim