source/src/agar/common/scaler/sse2/scaler_x6_sse2.c

   1 /*
   2  * Zoom x2x2
   3  * (C) 2013 K.Ohta
   4  *
   5  * History:
   6  *  2013-04-02 Move from scaler_x2.c
   7  */
   8 #include <agar/core.h>
   9 #include <agar/gui.h>
  10 #include "simd_types.h"
  11 #include "sdl_cpuid.h"
  12 #include "cache_wrapper.h"
  13
  14 extern struct XM7_CPUID *pCpuID;
  15 extern BOOL bFullScan;
  16
  17 extern void pVram2RGB_x6_Line(Uint32 *src, Uint32 *dst, int x, int y, int yrep);
  18
  19 #if defined(__SSE2__)
  20 static void Scaler_DrawLine(v4hi *dst, Uint32 *src, int ww, int repeat, int pitch)
  21 {
  22    int xx;
  23    int yy;
  24    int yrep2;
  25    int yrep3;
  26    int blank;
  27    v4hi *b2p;
  28    v4hi r1, r2;
  29    v4hi *d0;
  30    v4hi *b;
  31    int pitch2;
  32 #if AG_BIG_ENDIAN != 1
  33    const v4ui bb = {0xff000000, 0xff000000, 0xff000000, 0xff000000};
  34 #else
  35    const v4ui bb = {0x000000ff, 0x000000ff, 0x000000ff, 0x000000ff};
  36 #endif
  37
  38    if(repeat <= 0) return;
  39    b = (v4hi *)src;
  40    b2p = dst;
  41    pitch2 = pitch / sizeof(v4hi);
  42    if((bFullScan) || (repeat < 2)) {
  43       v4hi r3, r4, r5, r6, r7;
  44       v4hi r8, r9, r10, r11, r12;
  45       v4hi r13, r14;
  46       for(xx = 0; xx < ww; xx += 8) {
  47          b2p = dst;
  48          r1 = *b++;
  49          r2 = *b++;
  50          r3.uv   = (v4ui){r1.i[0], r1.i[0], r1.i[0], r1.i[0]};
  51          r4.uv   = (v4ui){r1.i[0], r1.i[0], r1.i[1], r1.i[1]};
  52          r5.uv   = (v4ui){r1.i[1], r1.i[1], r1.i[1], r1.i[1]};
  53          r6.uv   = (v4ui){r1.i[2], r1.i[2], r1.i[2], r1.i[2]};
  54          r7.uv   = (v4ui){r1.i[2], r1.i[2], r1.i[3], r1.i[3]};
  55          r8.uv   = (v4ui){r1.i[3], r1.i[3], r1.i[3], r1.i[3]};
  56
  57          r9.uv   = (v4ui){r2.i[0], r2.i[0], r2.i[0], r2.i[0]};
  58          r10.uv  = (v4ui){r2.i[0], r2.i[0], r2.i[1], r2.i[1]};
  59          r11.uv  = (v4ui){r2.i[1], r2.i[1], r2.i[1], r2.i[1]};
  60          r12.uv  = (v4ui){r2.i[2], r2.i[2], r2.i[2], r2.i[2]};
  61          r13.uv  = (v4ui){r2.i[2], r2.i[2], r2.i[3], r2.i[3]};
  62          r14.uv  = (v4ui){r2.i[3], r2.i[3], r2.i[3], r2.i[3]};
  63          for(yy = 0; yy < repeat; yy++) {
  64             b2p[0] = r3;
  65             b2p[1] = r4;
  66             b2p[2] = r5;
  67             b2p[3] = r6;
  68             b2p[4] = r7;
  69             b2p[5] = r8;
  70             b2p[6] = r9;
  71             b2p[7] = r10;
  72             b2p[8] = r11;
  73             b2p[9] = r12;
  74             b2p[10] = r13;
  75             b2p[11] = r14;
  76             b2p = b2p + pitch2;
  77          }
  78          dst += 10;
  79 //       b += 2;
  80       }
  81    } else {
  82       v4hi r3, r4, r5, r6, r7;
  83       v4hi r8, r9, r10, r11, r12;
  84       v4hi r13, r14;
  85       for(xx = 0; xx < ww; xx += 8) {
  86          b2p = dst;
  87          r1 = *b++;
  88          r2 = *b++;
  89
  90          r3.uv   = (v4ui){r1.i[0], r1.i[0], r1.i[0], r1.i[0]};
  91          r4.uv   = (v4ui){r1.i[0], r1.i[0], r1.i[1], r1.i[1]};
  92          r5.uv   = (v4ui){r1.i[1], r1.i[1], r1.i[1], r1.i[1]};
  93          r6.uv   = (v4ui){r1.i[2], r1.i[2], r1.i[2], r1.i[2]};
  94          r7.uv   = (v4ui){r1.i[2], r1.i[2], r1.i[3], r1.i[3]};
  95          r8.uv   = (v4ui){r1.i[3], r1.i[3], r1.i[3], r1.i[3]};
  96
  97          r9.uv   = (v4ui){r2.i[0], r2.i[0], r2.i[0], r2.i[0]};
  98          r10.uv  = (v4ui){r2.i[0], r2.i[0], r2.i[1], r2.i[1]};
  99          r11.uv  = (v4ui){r2.i[1], r2.i[1], r2.i[1], r2.i[1]};
 100          r12.uv  = (v4ui){r2.i[2], r2.i[2], r2.i[2], r2.i[2]};
 101          r13.uv  = (v4ui){r2.i[2], r2.i[2], r2.i[3], r2.i[3]};
 102          r14.uv  = (v4ui){r2.i[3], r2.i[3], r2.i[3], r2.i[3]};
 103          for(yy = 0; yy < repeat - 1; yy++) {
 104             b2p[0] = r3;
 105             b2p[1] = r4;
 106             b2p[2] = r5;
 107             b2p[3] = r6;
 108             b2p[4] = r7;
 109             b2p[5] = r8;
 110             b2p[6] = r9;
 111             b2p[7] = r10;
 112             b2p[8] = r11;
 113             b2p[9] = r12;
 114             b2p[10] = r13;
 115             b2p[11] = r14;
 116             b2p = b2p + pitch2;
 117          }
 118          b2p[0].uv =
 119          b2p[1].uv =
 120          b2p[2].uv =
 121          b2p[3].uv =
 122          b2p[4].uv =
 123          b2p[5].uv =
 124          b2p[6].uv =
 125          b2p[7].uv =
 126          b2p[8].uv =
 127          b2p[9].uv =
 128          b2p[10].uv =
 129          b2p[11].uv = bb;
 130          dst += 12;
 131 //       b += 2;
 132       }
 133    }
 134
 135 }
 136
 137
 138
 139 void pVram2RGB_x6_Line_SSE2(Uint32 *src, Uint8 *dst, int xbegin, int xend, int y, int yrep)
 140 {
 141    register v4hi *b;
 142    AG_Surface *Surface = GetDrawSurface();
 143    Uint32 *d1;
 144    Uint32 *d2;
 145    Uint32 *p;
 146    int w;
 147    int h;
 148    int yy;
 149    int xx;
 150    int hh;
 151    int ww;
 152    int i;
 153    int x = xbegin;
 154    int yrep2;
 155    unsigned  pitch;
 156    Uint32 black;
 157    if(Surface == NULL) return;
 158    w = Surface->w;
 159    h = Surface->h;
 160
 161
 162    ww = xend - xbegin;
 163 //   if(ww > (w / 2)) ww = w / 2;
 164    ww = (ww / 8) * 8;
 165    if(ww <= 0) return;
 166
 167
 168 #if AG_BIG_ENDIAN != 1
 169    black = 0xff000000;
 170 #else
 171    black = 0x000000ff;
 172 #endif
 173 //   yrep = yrep * 16.0f;
 174
 175    yrep2 = yrep;
 176
 177    d1 = (Uint32 *)((Uint8 *)dst + x * 6 * Surface->format->BytesPerPixel);
 178    d2 = &src[x + y * 640];
 179    Scaler_DrawLine((v4hi *)d1, (Uint32 *)d2, ww, yrep2, Surface->pitch);
 180 //   AG_SurfaceUnlock(Surface);
 181    return;
 182 }
 183
 184
 185 #else
 186
 187 void pVram2RGB_x6_Line_SSE2(Uint32 *src, int xbegin,  int xend, int y, int yrep)
 188 {
 189    pVram2RGB_x6_Line(src, dst, x, y, yrep);
 190 }
 191
 192 #endif // __SSE2__