[FFmpeg-devel] [PATCH 4/5] x86/opusdsp: implement FMA3 accelerated postfilter and deemphasis

Lynne dev at lynne.ee
Sat Mar 16 18:33:32 EET 2019


58893 decicycles in deemphasis_c,  130548 runs,    524 skips
9475 decicycles in deemphasis_fma3,  130686 runs,    386 skips -> 6.21x speedup

24866 decicycles in postfilter_c,   65386 runs,    150 skips
5268 decicycles in postfilter_fma3,   65505 runs,     31 skips -> 4.72x speedup

Total decoder speedup: ~14%

Deemphasis SIMD based on the following unrolling:
const float c1 = CELT_EMPH_COEFF, c2 = c1*c1, c3 = c2*c1, c4 = c3*c1;
float state = coeff;

for (int i = 0; i < len; i += 4) {
    y[0] = x[0] + c1*state;
    y[1] = x[1] + c2*state + c1*x[0];
    y[2] = x[2] + c3*state + c1*x[1] + c2*x[0];
    y[3] = x[3] + c4*state + c1*x[2] + c2*x[1] + c3*x[0];

    state = y[3];
    y += 4;
    x += 4;
}

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0004-x86-opusdsp-implement-FMA3-accelerated-postfilter-an.patch
Type: text/x-patch
Size: 8019 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20190316/b6d2f5f7/attachment.bin>


More information about the ffmpeg-devel mailing list