FFmpeg

94858 Commits

32 Branches

290MB

Tree: d620b1f9de

Commit Graph

Author	SHA1	Message	Date
Lynne	4d2f62150d	aarch64/opusdsp: implement NEON accelerated postfilter and deemphasis 153372 UNITS in postfilter_c, 65536 runs, 0 skips 73164 UNITS in postfilter_neon, 65536 runs, 0 skips -> 2.1x speedup 80591 UNITS in deemphasis_c, 131072 runs, 0 skips 43969 UNITS in deemphasis_neon, 131072 runs, 0 skips -> 1.83x speedup Total decoder speedup: ~15% on a Raspberry Pi 3 (from 28.1x to 33.5x realtime) Deemphasis SIMD based on the following unrolling: const float c1 = CELT_EMPH_COEFF, c2 = c1c1, c3 = c2c1, c4 = c3c1; float state = coeff; for (int i = 0; i < len; i += 4) { y[0] = x[0] + c1state; y[1] = x[1] + c2state + c1x[0]; y[2] = x[2] + c3state + c1x[1] + c2x[0]; y[3] = x[3] + c4state + c1x[2] + c2x[1] + c3*x[0]; state = y[3]; y += 4; x += 4; } Unlike the x86 version, duplication is used instead of pslldq so the structure and tables are different.	6 years ago
Lynne	605e330310	x86/opusdsp: implement FMA3 accelerated postfilter and deemphasis 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 = c1c1, c3 = c2c1, c4 = c3c1; float state = coeff; for (int i = 0; i < len; i += 4) { y[0] = x[0] + c1state; y[1] = x[1] + c2state + c1x[0]; y[2] = x[2] + c3state + c1x[1] + c2x[0]; y[3] = x[3] + c4state + c1x[2] + c2x[1] + c3*x[0]; state = y[3]; y += 4; x += 4; }	6 years ago
Lynne	0cea3ca894	opusdsp: create and move deemphasis and postfiltering from opus_celt	6 years ago

Author

SHA1

Message

Date

Lynne

4d2f62150d

aarch64/opusdsp: implement NEON accelerated postfilter and deemphasis

153372 UNITS in postfilter_c,   65536 runs,      0 skips
73164 UNITS in postfilter_neon,   65536 runs,      0 skips -> 2.1x speedup

80591 UNITS in deemphasis_c,  131072 runs,      0 skips
43969 UNITS in deemphasis_neon,  131072 runs,      0 skips -> 1.83x speedup

Total decoder speedup: ~15% on a Raspberry Pi 3 (from 28.1x to 33.5x realtime)

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;
}

Unlike the x86 version, duplication is used instead of pslldq so
the structure and tables are different.

6 years ago

Lynne

605e330310

x86/opusdsp: implement FMA3 accelerated postfilter and deemphasis

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;
}

6 years ago

Lynne

0cea3ca894

opusdsp: create and move deemphasis and postfiltering from opus_celt

6 years ago

3 Commits (d620b1f9de57573c8fb5b5035578d10c49df91a6)