I thought about including PSR but the article's take is still mostly accurate in that you do still send pixel by pixel, row by row, in the case of a panel with

I thought about including PSR but the article's take is still mostly accurate in that you do still send pixel by pixel, row by row, in the case of a panel with PSR. Just the panel then remembers those pixels. On something like an ARM SoC where this is all unified memory, I couldn't entirely remember where exactly PSR sat. In that case you could argue the transfer is an atomic switch to a different buffer, but I thought the PSR memory was still a different segment even in that case.