Closed GorgonMeducer closed 6 years ago
Closing this pull request - merged most of it, but some things are special:
Overall it improved the benchmark (running 1000 frames) by 2.7%. More to come...
Thanks! Just one thing, could I add macros to make the 16bit read interface and dedicated dma interface an optimisation option? I do see the improvements on my poor cortex-m platform. due to the very limited computing power, I have to do the cycle level optimisation.
I am doing next around update and will send a pull request again:
Cheers, Gabriel
Could you keep the pull request rather small - going optimisation by optimisation? It's easier for me to understand and merge it and we can discuss the pull requests separately.
I am very sorry about this. I will do it step by step in the future.