On Tue, 22 Apr 2014, Janne Grunau wrote:
Optimized for the default filter length 16.
30% faster opus silk decoding.
---
The assembler source is due to extensive macro use uglier than I would
like it to be but it's 4 functions in one. It looked nicer before I added
double support. The need for 4 neon registers for 8 doubles is the main
problem.
libavresample/aarch64/Makefile | 6 +-
libavresample/aarch64/resample_init.c | 64 ++++++++++
libavresample/aarch64/resample_neon.S | 219 ++++++++++++++++++++++++++++++++++
libavresample/internal.h | 3 +
libavresample/resample.c | 3 +
5 files changed, 293 insertions(+), 2 deletions(-)
create mode 100644 libavresample/aarch64/resample_init.c
create mode 100644 libavresample/aarch64/resample_neon.S
Looks ok to me
// Martin
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
