Closed tzneal closed 2 years ago
I split out the generic and SSE code into separate functions, this let me run benchmarks to make sure performance was improving.
Interested in your thoughts on the fast DC removal option I added, I went from a running average that's modified for every sample to one that's modified every 4 samples so I could vectorize it. The end result is a 20-30% speed improvement on the benchmark (and a corresponding 20-30% reduced CPU usage on the embedded board I'm playing with).
Thanks for that contribution and that nice improvement Do you have tested the build and checked all work fine on different platform ? (as I see there is a big refactoring)
@bvernoux I'm still doing a bit of testing, I'll probably verify it builds on Windows this week and I'll add a comment to let you know.
Do you have any update on this Pull Request ? (does it is finished/tested if yes could you add more details)
Closed "No feedback after more than 2 years" considered as cancelled