overall performance about froyvisuals-old HOT 2 CLOSED

I added profiling to libvisual.so See 'make profile' and gprof.sh.

It looks like depth transformation is the key bottleneck, and not VisDFT like I suspected. Another area of concern could be lv_math.c. That's a key component where NEON intrinsics may be useful on supported devices, as well as lv_video.c for depth transformations. We'll see. Header file arm_neon.h is massive.

from froyvisuals-old.

Performance is a lot better now that we shrunk the visualization size to 256x256 and just magnify that image with Canvas. We'll see about further improving, but it's really good now for the most part.

from froyvisuals-old.