1. ffmpeg
  2. Commit 605e3303109e7945a841c5fb5ff60881393933a3
Browse code

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;
}

Lynne authored on 2019/03/15 23:43:04
Showing 5 changed files
libavcodec/opusdsp.c
History View file @ 605e330
... ... 
@@ -58,4 +58,7 @@ av_cold void ff_opus_dsp_init(OpusDSP *ctx)
58 58 
 {
59 59 
     ctx->postfilter = postfilter_c;
60 60 
     ctx->deemphasis = deemphasis_c;
61 
+
62 
+    if (ARCH_X86)
63 
+        ff_opus_dsp_init_x86(ctx);
61 64 
 }
libavcodec/opusdsp.h
History View file @ 605e330
... ... 
@@ -30,4 +30,6 @@ typedef struct OpusDSP {
30 30
31 31 
 void ff_opus_dsp_init(OpusDSP *ctx);
32 32
33 
+void ff_opus_dsp_init_x86(OpusDSP *ctx);
34 
+
33 35 
 #endif /* AVCODEC_OPUSDSP_H */
libavcodec/x86/Makefile
History View file @ 605e330
... ... 
@@ -53,6 +53,7 @@ OBJS-$(CONFIG_CAVS_DECODER)            += x86/cavsdsp.o
53 53 
 OBJS-$(CONFIG_DCA_DECODER)             += x86/dcadsp_init.o x86/synth_filter_init.o
54 54 
 OBJS-$(CONFIG_DNXHD_ENCODER)           += x86/dnxhdenc_init.o
55 55 
 OBJS-$(CONFIG_EXR_DECODER)             += x86/exrdsp_init.o
56 
+OBJS-$(CONFIG_OPUS_DECODER)            += x86/opusdsp_init.o
56 57 
 OBJS-$(CONFIG_OPUS_ENCODER)            += x86/celt_pvq_init.o
57 58 
 OBJS-$(CONFIG_HEVC_DECODER)            += x86/hevcdsp_init.o
58 59 
 OBJS-$(CONFIG_JPEG2000_DECODER)        += x86/jpeg2000dsp_init.o
... ... 
@@ -126,6 +127,7 @@ X86ASM-OBJS-$(CONFIG_MDCT15)           += x86/mdct15.o
126 126 
 X86ASM-OBJS-$(CONFIG_ME_CMP)           += x86/me_cmp.o
127 127 
 X86ASM-OBJS-$(CONFIG_MPEGAUDIODSP)     += x86/imdct36.o
128 128 
 X86ASM-OBJS-$(CONFIG_MPEGVIDEOENC)     += x86/mpegvideoencdsp.o
129 
+X86ASM-OBJS-$(CONFIG_OPUS_DECODER)     += x86/opusdsp.o
129 130 
 X86ASM-OBJS-$(CONFIG_OPUS_ENCODER)     += x86/celt_pvq_search.o
130 131 
 X86ASM-OBJS-$(CONFIG_PIXBLOCKDSP)      += x86/pixblockdsp.o
131 132 
 X86ASM-OBJS-$(CONFIG_QPELDSP)          += x86/qpeldsp.o                 \
libavcodec/x86/opusdsp.asm
History View file @ 605e330
132 133 
new file mode 100644
... ... 
@@ -0,0 +1,114 @@
0 
+;******************************************************************************
1 
+;* Opus SIMD functions
2 
+;*
3 
+;* This file is part of FFmpeg.
4 
+;*
5 
+;* FFmpeg is free software; you can redistribute it and/or
6 
+;* modify it under the terms of the GNU Lesser General Public
7 
+;* License as published by the Free Software Foundation; either
8 
+;* version 2.1 of the License, or (at your option) any later version.
9 
+;*
10 
+;* FFmpeg is distributed in the hope that it will be useful,
11 
+;* but WITHOUT ANY WARRANTY; without even the implied warranty of
12 
+;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13 
+;* Lesser General Public License for more details.
14 
+;*
15 
+;* You should have received a copy of the GNU Lesser General Public
16 
+;* License along with FFmpeg; if not, write to the Free Software
17 
+;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18 
+;******************************************************************************
19 
+
20 
+%include "libavutil/x86/x86util.asm"
21 
+
22 
+SECTION_RODATA
23 
+
24 
+         ; 0.85..^1    0.85..^2    0.85..^3    0.85..^4
25 
+tab_st: dd 0x3f599a00, 0x3f38f671, 0x3f1d382a, 0x3f05a32f
26 
+tab_x0: dd 0x0,        0x3f599a00, 0x3f599a00, 0x3f599a00
27 
+tab_x1: dd 0x0,        0x0,        0x3f38f671, 0x3f38f671
28 
+tab_x2: dd 0x0,        0x0,        0x0,        0x3f1d382a
29 
+
30 
+SECTION .text
31 
+
32 
+INIT_XMM fma3
33 
+%if UNIX64
34 
+cglobal opus_deemphasis, 3, 3, 8, out, in, len
35 
+%else
36 
+cglobal opus_deemphasis, 4, 4, 8, out, in, coeff, len
37 
+%endif
38 
+%if ARCH_X86_32
39 
+    VBROADCASTSS m0, coeffm
40 
+%else
41 
+%if WIN64
42 
+    SWAP 0, 2
43 
+%endif
44 
+    shufps m0, m0, 0
45 
+%endif
46 
+
47 
+    movaps m4, [tab_st]
48 
+    movaps m5, [tab_x0]
49 
+    movaps m6, [tab_x1]
50 
+    movaps m7, [tab_x2]
51 
+
52 
+.loop:
53 
+    movaps  m1, [inq]                ; x0, x1, x2, x3
54 
+
55 
+    pslldq  m2, m1, 4                ;  0, x0, x1, x2
56 
+    pslldq  m3, m1, 8                ;  0,  0, x0, x1
57 
+
58 
+    fmaddps m2, m2, m5, m1           ; x + c1*x[0-2]
59 
+    pslldq  m1, 12                   ;  0,  0,  0, x0
60 
+
61 
+    fmaddps m2, m3, m6, m2           ; x + c1*x[0-2] + c2*x[0-1]
62 
+    fmaddps m1, m1, m7, m2           ; x + c1*x[0-2] + c2*x[0-1] + c3*x[0]
63 
+    fmaddps m0, m0, m4, m1           ; x + c1*x[0-2] + c2*x[0-1] + c3*x[0] + c*s
64 
+
65 
+    movaps [outq], m0
66 
+    shufps m0, m0, q3333             ; new state
67 
+
68 
+    add inq,  mmsize
69 
+    add outq, mmsize
70 
+    sub lenq, mmsize >> 2
71 
+    jg .loop
72 
+
73 
+%if ARCH_X86_64 == 0
74 
+    movss r0m, m0
75 
+    fld dword r0m
76 
+%endif
77 
+    RET
78 
+
79 
+
80 
+INIT_XMM fma3
81 
+cglobal opus_postfilter, 4, 4, 8, data, period, gains, len
82 
+    VBROADCASTSS m0, [gainsq + 0]
83 
+    VBROADCASTSS m1, [gainsq + 4]
84 
+    VBROADCASTSS m2, [gainsq + 8]
85 
+
86 
+    lea periodq, [periodq*4 + 8]
87 
+    neg periodq
88 
+
89 
+    movups  m3, [dataq + periodq]
90 
+    mulps   m3, m2
91 
+
92 
+.loop:
93 
+    movups  m4, [dataq + periodq +  4]
94 
+    movups  m5, [dataq + periodq +  8]
95 
+    movups  m6, [dataq + periodq + 12]
96 
+    movups  m7, [dataq + periodq + 16]
97 
+
98 
+    fmaddps m3, m7, m2, m3
99 
+    addps   m6, m4
100 
+
101 
+    fmaddps m5, m5, m0, [dataq]
102 
+    fmaddps m6, m6, m1, m3
103 
+
104 
+    addps   m5, m6
105 
+    mulps   m3, m7, m2
106 
+
107 
+    movaps  [dataq], m5
108 
+
109 
+    add dataq, mmsize
110 
+    sub lenq,  mmsize >> 2
111 
+    jg .loop
112 
+
113 
+    RET
libavcodec/x86/opusdsp_init.c
History View file @ 605e330
0 114 
new file mode 100644
... ... 
@@ -0,0 +1,35 @@
0 
+/*
1 
+ * This file is part of FFmpeg.
2 
+ *
3 
+ * FFmpeg is free software; you can redistribute it and/or
4 
+ * modify it under the terms of the GNU Lesser General Public
5 
+ * License as published by the Free Software Foundation; either
6 
+ * version 2.1 of the License, or (at your option) any later version.
7 
+ *
8 
+ * FFmpeg is distributed in the hope that it will be useful,
9 
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
11 
+ * Lesser General Public License for more details.
12 
+ *
13 
+ * You should have received a copy of the GNU Lesser General Public
14 
+ * License along with FFmpeg; if not, write to the Free Software
15 
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16 
+ */
17 
+
18 
+#include "config.h"
19 
+
20 
+#include "libavutil/x86/cpu.h"
21 
+#include "libavcodec/opusdsp.h"
22 
+
23 
+void ff_opus_postfilter_fma3(float *data, int period, float *gains, int len);
24 
+float ff_opus_deemphasis_fma3(float *out, float *in, float coeff, int len);
25 
+
26 
+av_cold void ff_opus_dsp_init_x86(OpusDSP *ctx)
27 
+{
28 
+    int cpu_flags = av_get_cpu_flags();
29 
+
30 
+    if (EXTERNAL_FMA3_FAST(cpu_flags)) {
31 
+        ctx->postfilter = ff_opus_postfilter_fma3;
32 
+        ctx->deemphasis = ff_opus_deemphasis_fma3;
33 
+    }
34 
+}