FFmpeg coverage

Directory:	../../../ffmpeg/
File:	src/libavcodec/faandct.c
Date:	2024-04-27 00:58:15

	Exec	Total	Coverage
Lines:	68	101	67.3%
Functions:	2	3	66.7%
Branches:	4	6	66.7%

  
      Line
      Branch
      Exec
      Source
    
      /*
    
       * Floating point AAN DCT
    
       * this implementation is based upon the IJG integer AAN DCT (see jfdctfst.c)
    
       *
    
       * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
    
       * Copyright (c) 2003 Roman Shaposhnik
    
       *
    
       * Permission to use, copy, modify, and/or distribute this software for any
    
       * purpose with or without fee is hereby granted, provided that the above
    
       * copyright notice and this permission notice appear in all copies.
    
       *
    
       * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
    
       * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
    
       * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
    
       * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
    
       * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
    
       * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
    
       * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
    
       */
    
      /**
    
       * @file
    
       * @brief
    
       *     Floating point AAN DCT
    
       * @author Michael Niedermayer <michaelni@gmx.at>
    
       */
    
      #include "faandct.h"
    
      #include "libavutil/emms.h"
    
      #include "libavutil/internal.h"
    
      #include "libavutil/libm.h"
    
      typedef float FLOAT;
    
      /* numbers generated by arbitrary precision arithmetic followed by truncation
    
      to 36 fractional digits (enough for a 128-bit IEEE quad, see /usr/include/math.h
    
      for this approach). Unfortunately, long double is not always available correctly,
    
      e.g ppc has issues.
    
      TODO: add L suffixes when ppc and toolchains sort out their stuff.
    
      */
    
      #define B0 1.000000000000000000000000000000000000
    
      #define B1 0.720959822006947913789091890943021267 // (cos(pi*1/16)sqrt(2))^-1
    
      #define B2 0.765366864730179543456919968060797734 // (cos(pi*2/16)sqrt(2))^-1
    
      #define B3 0.850430094767256448766702844371412325 // (cos(pi*3/16)sqrt(2))^-1
    
      #define B4 1.000000000000000000000000000000000000 // (cos(pi*4/16)sqrt(2))^-1
    
      #define B5 1.272758580572833938461007018281767032 // (cos(pi*5/16)sqrt(2))^-1
    
      #define B6 1.847759065022573512256366378793576574 // (cos(pi*6/16)sqrt(2))^-1
    
      #define B7 3.624509785411551372409941227504289587 // (cos(pi*7/16)sqrt(2))^-1
    
      #define A1 M_SQRT1_2              // cos(pi*4/16)
    
      #define A2 0.54119610014619698435 // cos(pi*6/16)sqrt(2)
    
      #define A5 0.38268343236508977170 // cos(pi*6/16)
    
      #define A4 1.30656296487637652774 // cos(pi*2/16)sqrt(2)
    
      static const FLOAT postscale[64]={
    
      B0*B0, B0*B1, B0*B2, B0*B3, B0*B4, B0*B5, B0*B6, B0*B7,
    
      B1*B0, B1*B1, B1*B2, B1*B3, B1*B4, B1*B5, B1*B6, B1*B7,
    
      B2*B0, B2*B1, B2*B2, B2*B3, B2*B4, B2*B5, B2*B6, B2*B7,
    
      B3*B0, B3*B1, B3*B2, B3*B3, B3*B4, B3*B5, B3*B6, B3*B7,
    
      B4*B0, B4*B1, B4*B2, B4*B3, B4*B4, B4*B5, B4*B6, B4*B7,
    
      B5*B0, B5*B1, B5*B2, B5*B3, B5*B4, B5*B5, B5*B6, B5*B7,
    
      B6*B0, B6*B1, B6*B2, B6*B3, B6*B4, B6*B5, B6*B6, B6*B7,
    
      B7*B0, B7*B1, B7*B2, B7*B3, B7*B4, B7*B5, B7*B6, B7*B7,
    
      };
    
      20000
      static av_always_inline void row_fdct(FLOAT temp[64], int16_t *data)
    
      {
    
          FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
    
          FLOAT tmp10, tmp11, tmp12, tmp13;
    
          FLOAT z2, z4, z11, z13;
    
          int i;
    
        2/2✓ Branch 0 taken 160000 times.
✓ Branch 1 taken 20000 times.

      180000
          for (i=0; i<8*8; i+=8) {
    
      160000
              tmp0= data[0 + i] + data[7 + i];
    
      160000
              tmp7= data[0 + i] - data[7 + i];
    
      160000
              tmp1= data[1 + i] + data[6 + i];
    
      160000
              tmp6= data[1 + i] - data[6 + i];
    
      160000
              tmp2= data[2 + i] + data[5 + i];
    
      160000
              tmp5= data[2 + i] - data[5 + i];
    
      160000
              tmp3= data[3 + i] + data[4 + i];
    
      160000
              tmp4= data[3 + i] - data[4 + i];
    
      160000
              tmp10= tmp0 + tmp3;
    
      160000
              tmp13= tmp0 - tmp3;
    
      160000
              tmp11= tmp1 + tmp2;
    
      160000
              tmp12= tmp1 - tmp2;
    
      160000
              temp[0 + i]= tmp10 + tmp11;
    
      160000
              temp[4 + i]= tmp10 - tmp11;
    
      160000
              tmp12 += tmp13;
    
      160000
              tmp12 *= A1;
    
      160000
              temp[2 + i]= tmp13 + tmp12;
    
      160000
              temp[6 + i]= tmp13 - tmp12;
    
      160000
              tmp4 += tmp5;
    
      160000
              tmp5 += tmp6;
    
      160000
              tmp6 += tmp7;
    
      160000
              z2= tmp4*(A2+A5) - tmp6*A5;
    
      160000
              z4= tmp6*(A4-A5) + tmp4*A5;
    
      160000
              tmp5*=A1;
    
      160000
              z11= tmp7 + tmp5;
    
      160000
              z13= tmp7 - tmp5;
    
      160000
              temp[5 + i]= z13 + z2;
    
      160000
              temp[3 + i]= z13 - z2;
    
      160000
              temp[1 + i]= z11 + z4;
    
      160000
              temp[7 + i]= z11 - z4;
    
          }
    
      20000
      }
    
      20000
      void ff_faandct(int16_t *data)
    
      {
    
          FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
    
          FLOAT tmp10, tmp11, tmp12, tmp13;
    
          FLOAT z2, z4, z11, z13;
    
          FLOAT temp[64];
    
          int i;
    
      20000
          emms_c();
    
      20000
          row_fdct(temp, data);
    
        2/2✓ Branch 0 taken 160000 times.
✓ Branch 1 taken 20000 times.

      180000
          for (i=0; i<8; i++) {
    
      160000
              tmp0= temp[8*0 + i] + temp[8*7 + i];
    
      160000
              tmp7= temp[8*0 + i] - temp[8*7 + i];
    
      160000
              tmp1= temp[8*1 + i] + temp[8*6 + i];
    
      160000
              tmp6= temp[8*1 + i] - temp[8*6 + i];
    
      160000
              tmp2= temp[8*2 + i] + temp[8*5 + i];
    
      160000
              tmp5= temp[8*2 + i] - temp[8*5 + i];
    
      160000
              tmp3= temp[8*3 + i] + temp[8*4 + i];
    
      160000
              tmp4= temp[8*3 + i] - temp[8*4 + i];
    
      160000
              tmp10= tmp0 + tmp3;
    
      160000
              tmp13= tmp0 - tmp3;
    
      160000
              tmp11= tmp1 + tmp2;
    
      160000
              tmp12= tmp1 - tmp2;
    
      160000
              data[8*0 + i]= lrintf(postscale[8*0 + i] * (tmp10 + tmp11));
    
      160000
              data[8*4 + i]= lrintf(postscale[8*4 + i] * (tmp10 - tmp11));
    
      160000
              tmp12 += tmp13;
    
      160000
              tmp12 *= A1;
    
      160000
              data[8*2 + i]= lrintf(postscale[8*2 + i] * (tmp13 + tmp12));
    
      160000
              data[8*6 + i]= lrintf(postscale[8*6 + i] * (tmp13 - tmp12));
    
      160000
              tmp4 += tmp5;
    
      160000
              tmp5 += tmp6;
    
      160000
              tmp6 += tmp7;
    
      160000
              z2= tmp4*(A2+A5) - tmp6*A5;
    
      160000
              z4= tmp6*(A4-A5) + tmp4*A5;
    
      160000
              tmp5*=A1;
    
      160000
              z11= tmp7 + tmp5;
    
      160000
              z13= tmp7 - tmp5;
    
      160000
              data[8*5 + i]= lrintf(postscale[8*5 + i] * (z13 + z2));
    
      160000
              data[8*3 + i]= lrintf(postscale[8*3 + i] * (z13 - z2));
    
      160000
              data[8*1 + i]= lrintf(postscale[8*1 + i] * (z11 + z4));
    
      160000
              data[8*7 + i]= lrintf(postscale[8*7 + i] * (z11 - z4));
    
          }
    
      20000
      }
    
      ✗
      void ff_faandct248(int16_t *data)
    
      {
    
          FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
    
          FLOAT tmp10, tmp11, tmp12, tmp13;
    
          FLOAT temp[64];
    
          int i;
    
      ✗
          emms_c();
    
      ✗
          row_fdct(temp, data);
    
      ✗
          for (i=0; i<8; i++) {
    
      ✗
              tmp0 = temp[8*0 + i] + temp[8*1 + i];
    
      ✗
              tmp1 = temp[8*2 + i] + temp[8*3 + i];
    
      ✗
              tmp2 = temp[8*4 + i] + temp[8*5 + i];
    
      ✗
              tmp3 = temp[8*6 + i] + temp[8*7 + i];
    
      ✗
              tmp4 = temp[8*0 + i] - temp[8*1 + i];
    
      ✗
              tmp5 = temp[8*2 + i] - temp[8*3 + i];
    
      ✗
              tmp6 = temp[8*4 + i] - temp[8*5 + i];
    
      ✗
              tmp7 = temp[8*6 + i] - temp[8*7 + i];
    
      ✗
              tmp10 = tmp0 + tmp3;
    
      ✗
              tmp11 = tmp1 + tmp2;
    
      ✗
              tmp12 = tmp1 - tmp2;
    
      ✗
              tmp13 = tmp0 - tmp3;
    
      ✗
              data[8*0 + i] = lrintf(postscale[8*0 + i] * (tmp10 + tmp11));
    
      ✗
              data[8*4 + i] = lrintf(postscale[8*4 + i] * (tmp10 - tmp11));
    
      ✗
              tmp12 += tmp13;
    
      ✗
              tmp12 *= A1;
    
      ✗
              data[8*2 + i] = lrintf(postscale[8*2 + i] * (tmp13 + tmp12));
    
      ✗
              data[8*6 + i] = lrintf(postscale[8*6 + i] * (tmp13 - tmp12));
    
      ✗
              tmp10 = tmp4 + tmp7;
    
      ✗
              tmp11 = tmp5 + tmp6;
    
      ✗
              tmp12 = tmp5 - tmp6;
    
      ✗
              tmp13 = tmp4 - tmp7;
    
      ✗
              data[8*1 + i] = lrintf(postscale[8*0 + i] * (tmp10 + tmp11));
    
      ✗
              data[8*5 + i] = lrintf(postscale[8*4 + i] * (tmp10 - tmp11));
    
      ✗
              tmp12 += tmp13;
    
      ✗
              tmp12 *= A1;
    
      ✗
              data[8*3 + i] = lrintf(postscale[8*2 + i] * (tmp13 + tmp12));
    
      ✗
              data[8*7 + i] = lrintf(postscale[8*6 + i] * (tmp13 - tmp12));
    
          }
    
      ✗
      }

Line	Branch	Exec	Source
1			/*
2			* Floating point AAN DCT
3			* this implementation is based upon the IJG integer AAN DCT (see jfdctfst.c)
4			*
5			* Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
6			* Copyright (c) 2003 Roman Shaposhnik
7			*
8			* Permission to use, copy, modify, and/or distribute this software for any
9			* purpose with or without fee is hereby granted, provided that the above
10			* copyright notice and this permission notice appear in all copies.
11			*
12			* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
13			* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
14			* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
15			* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
16			* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
17			* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
18			* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
19			*/
20
21			/**
22			* @file
23			* @brief
24			* Floating point AAN DCT
25			* @author Michael Niedermayer <michaelni@gmx.at>
26			*/
27
28			#include "faandct.h"
29			#include "libavutil/emms.h"
30			#include "libavutil/internal.h"
31			#include "libavutil/libm.h"
32
33			typedef float FLOAT;
34
35			/* numbers generated by arbitrary precision arithmetic followed by truncation
36			to 36 fractional digits (enough for a 128-bit IEEE quad, see /usr/include/math.h
37			for this approach). Unfortunately, long double is not always available correctly,
38			e.g ppc has issues.
39			TODO: add L suffixes when ppc and toolchains sort out their stuff.
40			*/
41			#define B0 1.000000000000000000000000000000000000
42			#define B1 0.720959822006947913789091890943021267 // (cos(pi*1/16)sqrt(2))^-1
43			#define B2 0.765366864730179543456919968060797734 // (cos(pi*2/16)sqrt(2))^-1
44			#define B3 0.850430094767256448766702844371412325 // (cos(pi*3/16)sqrt(2))^-1
45			#define B4 1.000000000000000000000000000000000000 // (cos(pi*4/16)sqrt(2))^-1
46			#define B5 1.272758580572833938461007018281767032 // (cos(pi*5/16)sqrt(2))^-1
47			#define B6 1.847759065022573512256366378793576574 // (cos(pi*6/16)sqrt(2))^-1
48			#define B7 3.624509785411551372409941227504289587 // (cos(pi*7/16)sqrt(2))^-1
49
50			#define A1 M_SQRT1_2 // cos(pi*4/16)
51			#define A2 0.54119610014619698435 // cos(pi*6/16)sqrt(2)
52			#define A5 0.38268343236508977170 // cos(pi*6/16)
53			#define A4 1.30656296487637652774 // cos(pi*2/16)sqrt(2)
54
55			static const FLOAT postscale[64]={
56			B0B0, B0B1, B0B2, B0B3, B0B4, B0B5, B0B6, B0B7,
57			B1B0, B1B1, B1B2, B1B3, B1B4, B1B5, B1B6, B1B7,
58			B2B0, B2B1, B2B2, B2B3, B2B4, B2B5, B2B6, B2B7,
59			B3B0, B3B1, B3B2, B3B3, B3B4, B3B5, B3B6, B3B7,
60			B4B0, B4B1, B4B2, B4B3, B4B4, B4B5, B4B6, B4B7,
61			B5B0, B5B1, B5B2, B5B3, B5B4, B5B5, B5B6, B5B7,
62			B6B0, B6B1, B6B2, B6B3, B6B4, B6B5, B6B6, B6B7,
63			B7B0, B7B1, B7B2, B7B3, B7B4, B7B5, B7B6, B7B7,
64			};
65
66		20000	static av_always_inline void row_fdct(FLOAT temp[64], int16_t *data)
67			{
68			FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
69			FLOAT tmp10, tmp11, tmp12, tmp13;
70			FLOAT z2, z4, z11, z13;
71			int i;
72
73	2/2 ✓ Branch 0 taken 160000 times. ✓ Branch 1 taken 20000 times.	180000	for (i=0; i<8*8; i+=8) {
74		160000	tmp0= data[0 + i] + data[7 + i];
75		160000	tmp7= data[0 + i] - data[7 + i];
76		160000	tmp1= data[1 + i] + data[6 + i];
77		160000	tmp6= data[1 + i] - data[6 + i];
78		160000	tmp2= data[2 + i] + data[5 + i];
79		160000	tmp5= data[2 + i] - data[5 + i];
80		160000	tmp3= data[3 + i] + data[4 + i];
81		160000	tmp4= data[3 + i] - data[4 + i];
82
83		160000	tmp10= tmp0 + tmp3;
84		160000	tmp13= tmp0 - tmp3;
85		160000	tmp11= tmp1 + tmp2;
86		160000	tmp12= tmp1 - tmp2;
87
88		160000	temp[0 + i]= tmp10 + tmp11;
89		160000	temp[4 + i]= tmp10 - tmp11;
90
91		160000	tmp12 += tmp13;
92		160000	tmp12 *= A1;
93		160000	temp[2 + i]= tmp13 + tmp12;
94		160000	temp[6 + i]= tmp13 - tmp12;
95
96		160000	tmp4 += tmp5;
97		160000	tmp5 += tmp6;
98		160000	tmp6 += tmp7;
99
100		160000	z2= tmp4(A2+A5) - tmp6A5;
101		160000	z4= tmp6(A4-A5) + tmp4A5;
102
103		160000	tmp5*=A1;
104
105		160000	z11= tmp7 + tmp5;
106		160000	z13= tmp7 - tmp5;
107
108		160000	temp[5 + i]= z13 + z2;
109		160000	temp[3 + i]= z13 - z2;
110		160000	temp[1 + i]= z11 + z4;
111		160000	temp[7 + i]= z11 - z4;
112			}
113		20000	}
114
115		20000	void ff_faandct(int16_t *data)
116			{
117			FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
118			FLOAT tmp10, tmp11, tmp12, tmp13;
119			FLOAT z2, z4, z11, z13;
120			FLOAT temp[64];
121			int i;
122
123		20000	emms_c();
124
125		20000	row_fdct(temp, data);
126
127	2/2 ✓ Branch 0 taken 160000 times. ✓ Branch 1 taken 20000 times.	180000	for (i=0; i<8; i++) {
128		160000	tmp0= temp[80 + i] + temp[87 + i];
129		160000	tmp7= temp[80 + i] - temp[87 + i];
130		160000	tmp1= temp[81 + i] + temp[86 + i];
131		160000	tmp6= temp[81 + i] - temp[86 + i];
132		160000	tmp2= temp[82 + i] + temp[85 + i];
133		160000	tmp5= temp[82 + i] - temp[85 + i];
134		160000	tmp3= temp[83 + i] + temp[84 + i];
135		160000	tmp4= temp[83 + i] - temp[84 + i];
136
137		160000	tmp10= tmp0 + tmp3;
138		160000	tmp13= tmp0 - tmp3;
139		160000	tmp11= tmp1 + tmp2;
140		160000	tmp12= tmp1 - tmp2;
141
142		160000	data[80 + i]= lrintf(postscale[80 + i] * (tmp10 + tmp11));
143		160000	data[84 + i]= lrintf(postscale[84 + i] * (tmp10 - tmp11));
144
145		160000	tmp12 += tmp13;
146		160000	tmp12 *= A1;
147		160000	data[82 + i]= lrintf(postscale[82 + i] * (tmp13 + tmp12));
148		160000	data[86 + i]= lrintf(postscale[86 + i] * (tmp13 - tmp12));
149
150		160000	tmp4 += tmp5;
151		160000	tmp5 += tmp6;
152		160000	tmp6 += tmp7;
153
154		160000	z2= tmp4(A2+A5) - tmp6A5;
155		160000	z4= tmp6(A4-A5) + tmp4A5;
156
157		160000	tmp5*=A1;
158
159		160000	z11= tmp7 + tmp5;
160		160000	z13= tmp7 - tmp5;
161
162		160000	data[85 + i]= lrintf(postscale[85 + i] * (z13 + z2));
163		160000	data[83 + i]= lrintf(postscale[83 + i] * (z13 - z2));
164		160000	data[81 + i]= lrintf(postscale[81 + i] * (z11 + z4));
165		160000	data[87 + i]= lrintf(postscale[87 + i] * (z11 - z4));
166			}
167		20000	}
168
169		✗	void ff_faandct248(int16_t *data)
170			{
171			FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
172			FLOAT tmp10, tmp11, tmp12, tmp13;
173			FLOAT temp[64];
174			int i;
175
176		✗	emms_c();
177
178		✗	row_fdct(temp, data);
179
180		✗	for (i=0; i<8; i++) {
181		✗	tmp0 = temp[80 + i] + temp[81 + i];
182		✗	tmp1 = temp[82 + i] + temp[83 + i];
183		✗	tmp2 = temp[84 + i] + temp[85 + i];
184		✗	tmp3 = temp[86 + i] + temp[87 + i];
185		✗	tmp4 = temp[80 + i] - temp[81 + i];
186		✗	tmp5 = temp[82 + i] - temp[83 + i];
187		✗	tmp6 = temp[84 + i] - temp[85 + i];
188		✗	tmp7 = temp[86 + i] - temp[87 + i];
189
190		✗	tmp10 = tmp0 + tmp3;
191		✗	tmp11 = tmp1 + tmp2;
192		✗	tmp12 = tmp1 - tmp2;
193		✗	tmp13 = tmp0 - tmp3;
194
195		✗	data[80 + i] = lrintf(postscale[80 + i] * (tmp10 + tmp11));
196		✗	data[84 + i] = lrintf(postscale[84 + i] * (tmp10 - tmp11));
197
198		✗	tmp12 += tmp13;
199		✗	tmp12 *= A1;
200		✗	data[82 + i] = lrintf(postscale[82 + i] * (tmp13 + tmp12));
201		✗	data[86 + i] = lrintf(postscale[86 + i] * (tmp13 - tmp12));
202
203		✗	tmp10 = tmp4 + tmp7;
204		✗	tmp11 = tmp5 + tmp6;
205		✗	tmp12 = tmp5 - tmp6;
206		✗	tmp13 = tmp4 - tmp7;
207
208		✗	data[81 + i] = lrintf(postscale[80 + i] * (tmp10 + tmp11));
209		✗	data[85 + i] = lrintf(postscale[84 + i] * (tmp10 - tmp11));
210
211		✗	tmp12 += tmp13;
212		✗	tmp12 *= A1;
213		✗	data[83 + i] = lrintf(postscale[82 + i] * (tmp13 + tmp12));
214		✗	data[87 + i] = lrintf(postscale[86 + i] * (tmp13 - tmp12));
215			}
216		✗	}
217

Function (Line)	Call count	Block coverage
ff_faandct (line 115)	called 20000 times, returned 20000 times	100.0%
ff_faandct248 (line 169)	not called	0.0%
row_fdct (line 66)	called 20000 times, returned 20000 times	100.0%