kwave currently ships with optimised/accelerated implementations of memcpy for various architectures, sourced from xine. This shouldn't be needed these days, as glibc at least has various SIMD memcpy implementations, but also, compilers are very good at both pattern recognition and vectorisation these days.
A possibly relevant merge request was started @ https://invent.kde.org/multimedia/kwave/-/merge_requests/59
Git commit fbff057976e8988e01de0430705a8c75573bcd29 by Mark Penner. Committed on 11/02/2025 at 00:23. Pushed by mrp into branch 'master'. use standard memcpy the old custom memcpy implementations don't compile correctly with some compilers and are not usually faster than the modern standard memcpy anyway Related: bug 487934 M +1 -4 .flatpak-manifest.json M +0 -6 AUTHORS M +0 -40 CMakeLists.txt M +0 -3 LICENSES M +0 -15 config.h.cmake M +0 -7 doc/devel.docbook M +0 -14 doc/en/index.docbook M +0 -1 doxy.cfg.in M +0 -11 kwave/main.cpp M +0 -37 libkwave/CMakeLists.txt M +0 -1 libkwave/MultiPlaybackSink.cpp M +0 -1 libkwave/TrackWriter.cpp D +0 -429 libkwave/cpu_accel.cpp D +0 -53 libkwave/cputest.h D +0 -702 libkwave/memcpy.c M +0 -14 libkwave/memcpy.h https://invent.kde.org/multimedia/kwave/-/commit/fbff057976e8988e01de0430705a8c75573bcd29