Context: Previously I used the CPU for rendering pixelart graphics, but then I realized I could not only render to a low-res texture that I then upscale, but also write a shader to do color lookup.
Context: Previously I used the CPU for rendering pixelart graphics, but then I realized I could not only render to a low-res texture that I then upscale, but also write a shader to do color lookup.
These are all CPU rendered when they were made, but this is how the benchmark looks on the Raspberry Pi 400 (heavily handicapped by the memory bandwidth):
GUI example, with real bitmap fonts and CPU drawn windows:
GUI drawing algorithms will remain on the CPU, then I just upstream any changes finalized to the GPU.