--- /dev/null
+/*****************************************************************************\r
+ *\r
+ * XVID MPEG-4 VIDEO CODEC\r
+ * - Inverse DCT -\r
+ *\r
+ * Copyright (C) 2006-2011 Xvid Solutions GmbH\r
+ *\r
+ * This program is free software ; you can redistribute it and/or modify\r
+ * it under the terms of the GNU General Public License as published by\r
+ * the Free Software Foundation ; either version 2 of the License, or\r
+ * (at your option) any later version.\r
+ *\r
+ * This program is distributed in the hope that it will be useful,\r
+ * but WITHOUT ANY WARRANTY ; without even the implied warranty of\r
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\r
+ * GNU General Public License for more details.\r
+ *\r
+ * You should have received a copy of the GNU General Public License\r
+ * along with this program ; if not, write to the Free Software\r
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA\r
+ *\r
+ * $Id: idct.c 1986 2011-05-18 09:07:40Z Isibaar $\r
+ *\r
+ ****************************************************************************/\r
+\r
+/*\r
+ * Authors: Skal\r
+ *\r
+ * Walken IDCT\r
+ * Alternative idct implementations for decoding compatibility\r
+ *\r
+ * NOTE: this "C" version is not the original one,\r
+ * but is modified to yield the same error profile\r
+ * than the MMX version.\r
+ *\r
+ ************************************************************************/\r
+\r
+#include <stdint.h>\r
+\r
+//#include "idct.h"\r
+\r
+/* function pointer */\r
+//idctFuncPtr idct;\r
+\r
+#define XVID_DSP_CLIP_255(x) ( ((x)&~255) ? ((-(x)) >> (8*sizeof((x))-1))&0xff : (x) )\r
+\r
+#define ROW_SHIFT 11\r
+#define COL_SHIFT 6\r
+\r
+// #define FIX(x) (int)((x) * (1<<ROW_SHIFT))\r
+#define Rnd0 65536 // 1<<(COL_SHIFT+ROW_SHIFT-1);\r
+#define Rnd1 3597 // FIX (1.75683487303);\r
+#define Rnd2 2260 // FIX (1.10355339059);\r
+#define Rnd3 1203 // FIX (0.587788325588);\r
+#define Rnd4 0\r
+#define Rnd5 120 // FIX (0.058658283817);\r
+#define Rnd6 512 // FIX (0.25);\r
+#define Rnd7 512 // FIX (0.25);\r
+#undef FIX\r
+\r
+static const int Tab04[] = { 22725, 21407, 19266, 16384, 12873, 8867, 4520 };\r
+static const int Tab17[] = { 31521, 29692, 26722, 22725, 17855, 12299, 6270 };\r
+static const int Tab26[] = { 29692, 27969, 25172, 21407, 16819, 11585, 5906 };\r
+static const int Tab35[] = { 26722, 25172, 22654, 19266, 15137, 10426, 5315 };\r
+\r
+static int Idct_Row(short * In, const int * const Tab, int Rnd)\r
+{\r
+ const int C1 = Tab[0];\r
+ const int C2 = Tab[1];\r
+ const int C3 = Tab[2];\r
+ const int C4 = Tab[3];\r
+ const int C5 = Tab[4];\r
+ const int C6 = Tab[5];\r
+ const int C7 = Tab[6];\r
+\r
+ const int Right = In[5]|In[6]|In[7];\r
+ const int Left = In[1]|In[2]|In[3];\r
+ if (!(Right | In[4]))\r
+ {\r
+ const int K = C4*In[0] + Rnd;\r
+ if (Left)\r
+ {\r
+ const int a0 = K + C2*In[2];\r
+ const int a1 = K + C6*In[2];\r
+ const int a2 = K - C6*In[2];\r
+ const int a3 = K - C2*In[2];\r
+\r
+ const int b0 = C1*In[1] + C3*In[3];\r
+ const int b1 = C3*In[1] - C7*In[3];\r
+ const int b2 = C5*In[1] - C1*In[3];\r
+ const int b3 = C7*In[1] - C5*In[3];\r
+\r
+ In[0] = (a0 + b0) >> ROW_SHIFT;\r
+ In[1] = (a1 + b1) >> ROW_SHIFT;\r
+ In[2] = (a2 + b2) >> ROW_SHIFT;\r
+ In[3] = (a3 + b3) >> ROW_SHIFT;\r
+ In[4] = (a3 - b3) >> ROW_SHIFT;\r
+ In[5] = (a2 - b2) >> ROW_SHIFT;\r
+ In[6] = (a1 - b1) >> ROW_SHIFT;\r
+ In[7] = (a0 - b0) >> ROW_SHIFT;\r
+ }\r
+ else\r
+ {\r
+ const int a0 = K >> ROW_SHIFT;\r
+ if (a0) {\r
+ In[0] = In[1] = In[2] = In[3] =\r
+ In[4] = In[5] = In[6] = In[7] = a0;\r
+ }\r
+ else return 0;\r
+ }\r
+ }\r
+ else if (!(Left|Right))\r
+ {\r
+ const int a0 = (Rnd + C4*(In[0]+In[4])) >> ROW_SHIFT;\r
+ const int a1 = (Rnd + C4*(In[0]-In[4])) >> ROW_SHIFT;\r
+\r
+ In[0] = a0;\r
+ In[3] = a0;\r
+ In[4] = a0;\r
+ In[7] = a0;\r
+ In[1] = a1;\r
+ In[2] = a1;\r
+ In[5] = a1;\r
+ In[6] = a1;\r
+ }\r
+ else\r
+ {\r
+ const int K = C4*In[0] + Rnd;\r
+ const int a0 = K + C2*In[2] + C4*In[4] + C6*In[6];\r
+ const int a1 = K + C6*In[2] - C4*In[4] - C2*In[6];\r
+ const int a2 = K - C6*In[2] - C4*In[4] + C2*In[6];\r
+ const int a3 = K - C2*In[2] + C4*In[4] - C6*In[6];\r
+\r
+ const int b0 = C1*In[1] + C3*In[3] + C5*In[5] + C7*In[7];\r
+ const int b1 = C3*In[1] - C7*In[3] - C1*In[5] - C5*In[7];\r
+ const int b2 = C5*In[1] - C1*In[3] + C7*In[5] + C3*In[7];\r
+ const int b3 = C7*In[1] - C5*In[3] + C3*In[5] - C1*In[7];\r
+\r
+ In[0] = (a0 + b0) >> ROW_SHIFT;\r
+ In[1] = (a1 + b1) >> ROW_SHIFT;\r
+ In[2] = (a2 + b2) >> ROW_SHIFT;\r
+ In[3] = (a3 + b3) >> ROW_SHIFT;\r
+ In[4] = (a3 - b3) >> ROW_SHIFT;\r
+ In[5] = (a2 - b2) >> ROW_SHIFT;\r
+ In[6] = (a1 - b1) >> ROW_SHIFT;\r
+ In[7] = (a0 - b0) >> ROW_SHIFT;\r
+ }\r
+ return 1;\r
+}\r
+\r
+#define Tan1 0x32ec\r
+#define Tan2 0x6a0a\r
+#define Tan3 0xab0e\r
+#define Sqrt2 0x5a82\r
+\r
+#define MULT(c,x, n) ( ((c) * (x)) >> (n) )\r
+// 12b version => #define MULT(c,x, n) ( (((c)>>3) * (x)) >> ((n)-3) )\r
+// 12b zero-testing version:\r
+\r
+#define BUTF(a, b, tmp) \\r
+ (tmp) = (a)+(b); \\r
+ (b) = (a)-(b); \\r
+ (a) = (tmp)\r
+\r
+#define LOAD_BUTF(m1, m2, a, b, tmp, S) \\r
+ (m1) = (S)[(a)] + (S)[(b)]; \\r
+ (m2) = (S)[(a)] - (S)[(b)]\r
+\r
+static void Idct_Col_8(short * const In)\r
+{\r
+ int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, Spill;\r
+\r
+ // odd\r
+\r
+ mm4 = (int)In[7*8];\r
+ mm5 = (int)In[5*8];\r
+ mm6 = (int)In[3*8];\r
+ mm7 = (int)In[1*8];\r
+\r
+ mm0 = MULT(Tan1, mm4, 16) + mm7;\r
+ mm1 = MULT(Tan1, mm7, 16) - mm4;\r
+ mm2 = MULT(Tan3, mm5, 16) + mm6;\r
+ mm3 = MULT(Tan3, mm6, 16) - mm5;\r
+\r
+ mm7 = mm0 + mm2;\r
+ mm4 = mm1 - mm3;\r
+ mm0 = mm0 - mm2;\r
+ mm1 = mm1 + mm3;\r
+ mm6 = mm0 + mm1;\r
+ mm5 = mm0 - mm1;\r
+ mm5 = 2*MULT(Sqrt2, mm5, 16); // 2*sqrt2\r
+ mm6 = 2*MULT(Sqrt2, mm6, 16); // Watch out: precision loss but done to match\r
+ // the pmulhw used in mmx/sse versions\r
+ \r
+ // even\r
+\r
+ mm1 = (int)In[2*8];\r
+ mm2 = (int)In[6*8];\r
+ mm3 = MULT(Tan2,mm2, 16) + mm1;\r
+ mm2 = MULT(Tan2,mm1, 16) - mm2;\r
+\r
+ LOAD_BUTF(mm0, mm1, 0*8, 4*8, Spill, In);\r
+\r
+ BUTF(mm0, mm3, Spill);\r
+ BUTF(mm0, mm7, Spill);\r
+ In[8*0] = (int16_t) (mm0 >> COL_SHIFT);\r
+ In[8*7] = (int16_t) (mm7 >> COL_SHIFT);\r
+ BUTF(mm3, mm4, mm0);\r
+ In[8*3] = (int16_t) (mm3 >> COL_SHIFT);\r
+ In[8*4] = (int16_t) (mm4 >> COL_SHIFT);\r
+\r
+ BUTF(mm1, mm2, mm0);\r
+ BUTF(mm1, mm6, mm0);\r
+ In[8*1] = (int16_t) (mm1 >> COL_SHIFT);\r
+ In[8*6] = (int16_t) (mm6 >> COL_SHIFT);\r
+ BUTF(mm2, mm5, mm0);\r
+ In[8*2] = (int16_t) (mm2 >> COL_SHIFT);\r
+ In[8*5] = (int16_t) (mm5 >> COL_SHIFT);\r
+}\r
+\r
+static void Idct_Col_4(short * const In)\r
+{\r
+ int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, Spill;\r
+\r
+ // odd\r
+\r
+ mm0 = (int)In[1*8];\r
+ mm2 = (int)In[3*8];\r
+\r
+ mm1 = MULT(Tan1, mm0, 16);\r
+ mm3 = MULT(Tan3, mm2, 16);\r
+\r
+ mm7 = mm0 + mm2;\r
+ mm4 = mm1 - mm3;\r
+ mm0 = mm0 - mm2;\r
+ mm1 = mm1 + mm3;\r
+ mm6 = mm0 + mm1;\r
+ mm5 = mm0 - mm1;\r
+ mm6 = 2*MULT(Sqrt2, mm6, 16); // 2*sqrt2\r
+ mm5 = 2*MULT(Sqrt2, mm5, 16);\r
+\r
+ // even\r
+\r
+ mm0 = mm1 = (int)In[0*8];\r
+ mm3 = (int)In[2*8];\r
+ mm2 = MULT(Tan2,mm3, 16);\r
+\r
+ BUTF(mm0, mm3, Spill);\r
+ BUTF(mm0, mm7, Spill);\r
+ In[8*0] = (int16_t) (mm0 >> COL_SHIFT);\r
+ In[8*7] = (int16_t) (mm7 >> COL_SHIFT);\r
+ BUTF(mm3, mm4, mm0);\r
+ In[8*3] = (int16_t) (mm3 >> COL_SHIFT);\r
+ In[8*4] = (int16_t) (mm4 >> COL_SHIFT);\r
+\r
+ BUTF(mm1, mm2, mm0);\r
+ BUTF(mm1, mm6, mm0);\r
+ In[8*1] = (int16_t) (mm1 >> COL_SHIFT);\r
+ In[8*6] = (int16_t) (mm6 >> COL_SHIFT);\r
+ BUTF(mm2, mm5, mm0);\r
+ In[8*2] = (int16_t) (mm2 >> COL_SHIFT);\r
+ In[8*5] = (int16_t) (mm5 >> COL_SHIFT);\r
+}\r
+\r
+static void Idct_Col_3(short * const In)\r
+{\r
+ int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, Spill;\r
+\r
+ // odd\r
+\r
+ mm7 = (int)In[1*8];\r
+ mm4 = MULT(Tan1, mm7, 16);\r
+\r
+ mm6 = mm7 + mm4;\r
+ mm5 = mm7 - mm4;\r
+ mm6 = 2*MULT(Sqrt2, mm6, 16); // 2*sqrt2\r
+ mm5 = 2*MULT(Sqrt2, mm5, 16);\r
+\r
+ // even\r
+\r
+ mm0 = mm1 = (int)In[0*8];\r
+ mm3 = (int)In[2*8];\r
+ mm2 = MULT(Tan2,mm3, 16);\r
+\r
+ BUTF(mm0, mm3, Spill);\r
+ BUTF(mm0, mm7, Spill);\r
+ In[8*0] = (int16_t) (mm0 >> COL_SHIFT);\r
+ In[8*7] = (int16_t) (mm7 >> COL_SHIFT);\r
+ BUTF(mm3, mm4, mm0);\r
+ In[8*3] = (int16_t) (mm3 >> COL_SHIFT);\r
+ In[8*4] = (int16_t) (mm4 >> COL_SHIFT);\r
+\r
+ BUTF(mm1, mm2, mm0);\r
+ BUTF(mm1, mm6, mm0);\r
+ In[8*1] = (int16_t) (mm1 >> COL_SHIFT);\r
+ In[8*6] = (int16_t) (mm6 >> COL_SHIFT);\r
+ BUTF(mm2, mm5, mm0);\r
+ In[8*2] = (int16_t) (mm2 >> COL_SHIFT);\r
+ In[8*5] = (int16_t) (mm5 >> COL_SHIFT);\r
+}\r
+\r
+#undef Tan1\r
+#undef Tan2\r
+#undef Tan3\r
+#undef Sqrt2\r
+\r
+#undef ROW_SHIFT\r
+#undef COL_SHIFT\r
+\r
+//////////////////////////////////////////////////////////\r
+\r
+void idct_int32(short *const In)\r
+{\r
+ int i, Rows = 0x07;\r
+\r
+ Idct_Row(In + 0*8, Tab04, Rnd0);\r
+ Idct_Row(In + 1*8, Tab17, Rnd1);\r
+ Idct_Row(In + 2*8, Tab26, Rnd2);\r
+ if (Idct_Row(In + 3*8, Tab35, Rnd3)) Rows |= 0x08;\r
+ if (Idct_Row(In + 4*8, Tab04, Rnd4)) Rows |= 0x10;\r
+ if (Idct_Row(In + 5*8, Tab35, Rnd5)) Rows |= 0x20;\r
+ if (Idct_Row(In + 6*8, Tab26, Rnd6)) Rows |= 0x40;\r
+ if (Idct_Row(In + 7*8, Tab17, Rnd7)) Rows |= 0x80;\r
+\r
+ if (Rows&0xf0) {\r
+ for(i=0; i<8; i++)\r
+ Idct_Col_8(In + i);\r
+ }\r
+ else if (Rows&0x08) {\r
+ for(i=0; i<8; i++)\r
+ Idct_Col_4(In + i);\r
+ }\r
+ else {\r
+ for(i=0; i<8; i++)\r
+ Idct_Col_3(In + i);\r
+ }\r
+}\r
projects / narabu / blobdiff