audio-resampler: optimizations
authorWim Taymans <wtaymans@redhat.com>
Wed, 13 Jan 2016 16:44:39 +0000 (17:44 +0100)
committerWim Taymans <wtaymans@redhat.com>
Mon, 28 Mar 2016 11:25:50 +0000 (13:25 +0200)
commit71871c50483df00cd46dcfcb99148a4cd0097dfc
tree8933f8dedd7c8f7f48f09f14ac4fc7471bc729aa
parentf55a67ca7cbded75b383791c4d0ca52993524f55
audio-resampler: optimizations

Improve int16 resampling by using pmaddwd
Use intrinsics to scale and pack int16 samples
Align the coefficients so that we can use aligned loads
Add padding to taps and samples so that we don't have to use partial
loads for the remainder of the loops.
Remove copy_n, we can reuse the plain copy function with some new
parameters.
Align and pad the sample array.
gst-libs/gst/audio/audio-resampler-x86.h
gst-libs/gst/audio/audio-resampler.c