Improve YUV to RGB Performance II

[jnnks]

Rewrites the YUV 2 RGB Loop

• cache u & v planes for two rows of RGB pixels
• use addition to increase indices

Before:

test convert_yuv_to_rgb_512x512    ... bench:     532,512.85 ns/iter (+/- 75,759.37)

After:

test convert_yuv_to_rgb_512x512    ... bench:     444,830.35 ns/iter (+/- 38,899.95)

[ralfbiedert]

At this rate, by the end of the year you're beating GPUs :)

Outstanding, thanks!

[ralfbiedert]

Can confirm numbers, went from 191μs to 169μs.

[jnnks]

At this rate, by the end of the year you're beating GPUs :)

I'm not so sure about that :D

That was fun. I had to do this conversion so often already, it was about time to do it a bit more sophisticated 👍