summaryrefslogtreecommitdiff
path: root/src/mesa/pipe/llvm/llvm_builtins.c
blob: e135b4781d618b0140e8b51cc58c9b4abfffb85d (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
/* clang --emit-llvm llvm_builtins.c |llvm-as |opt -std-compile-opts |llvm-dis */
/* clang --emit-llvm llvm_builtins.c |llvm-as |opt -std-compile-opts |llvm2cpp -for=Shader -gen-module -funcname=createBaseShader */
typedef __attribute__(( ocu_vector_type(4) )) float float4;

#if 0
//clang doesn't suppoer "struct->member" notation yet
struct vertex_header {
   unsigned clipmask:12;
   unsigned edgeflag:1;
   unsigned pad:3;
   unsigned vertex_id:16;

   float clip[4];

   float data[][4];
};

inline float
dot4(float4 a, float4 b)
{
   float4 c = a*b;
   return c.x + c.y + c.z + c.w;
}

inline unsigned
compute_clipmask(float4 clip, float4 (*plane), unsigned nr)
{
   unsigned mask = 0;
   unsigned i;

   for (i = 0; i < nr; i++) {
      if (dot4(clip, plane[i]) < 0)
         mask |= (1<<i);
   }

   return mask;
}

inline void collect_results(float4 *results, struct vertex_header *vOut,
                            float4 *planes, int nr_planes,
                            float4 scale, float4 trans,
                            int num_attribs)
{
   /* store results */
   unsigned slot;
   float x, y, z, w;

   /* Handle attr[0] (position) specially:
    */
   float4 res0 = results[0];
   float *clip = vOut->clip;
   x = clip[0] = res0.x;
   y = clip[1] = res0.y;
   z = clip[2] = res0.z;
   w = clip[3] = res0.w;

   vOut->clipmask = compute_clipmask(res0, planes, nr_planes);
   vOut->edgeflag = 1;

   /* divide by w */
   w = 1.0f / w;
   x *= w;
   y *= w;
   z *= w;
   res0.x = x; res0.y = y; res0.z = z; res0.w = 1;

   /* Viewport mapping */
   res0 = res0 * scale + trans;
   vOut->data[0][0] = res0.x;
   vOut->data[0][1] = res0.y;
   vOut->data[0][2] = res0.z;
   vOut->data[0][3] = w;

   /* Remaining attributes are packed into sequential post-transform
    * vertex attrib slots.
    * Skip 0 since we just did it above.
    * Subtract two because of the VERTEX_HEADER, CLIP_POS attribs.
    */
   for (slot = 1; slot < num_attribs - 2; slot++) {
      float4 vec = results[slot];
      vOut->data[slot][0] = vec.x;
      vOut->data[slot][1] = vec.y;
      vOut->data[slot][2] = vec.z;
      vOut->data[slot][3] = vec.w;

      printf("output %d: %f %f %f %f\n", slot,
             vOut->data[slot][0],
             vOut->data[slot][1],
             vOut->data[slot][2],
             vOut->data[slot][3]);
   }
}
#endif

void from_array(float4 (*res)[16], float (*ainputs)[16][4],
                int count, int num_attribs)
{
   for (int i = 0; i < count; ++i) {
      for (int j = 0; j < num_attribs; ++j) {
         float4 vec;
         vec.x = ainputs[i][j][0];
         vec.y = ainputs[i][j][1];
         vec.z = ainputs[i][j][2];
         vec.w = ainputs[i][j][3];
         //printf("FAR(%d %d) %f %f %f %f\n", i, j, vec.x, vec.y, vec.z, vec.w);
         res[i][j] = vec;
      }
   }
}

void from_consts(float4 *res, float (*ainputs)[4],
                int count)
{
   for (int i = 0; i < count; ++i) {
      float4 vec;
      vec.x = ainputs[i][0];
      vec.y = ainputs[i][1];
      vec.z = ainputs[i][2];
      vec.w = ainputs[i][3];
      //printf("VCONST %f %f %f %f\n", vec.x, vec.y, vec.z, vec.w);
      res[i] = vec;
   }
}

void to_array(float (*dests)[4], float4 *in, int num_attribs)
{
   for (int i = 0; i < num_attribs; ++i) {
      float  *rd = dests[i];
      float4  ri = in[i];
      printf("DEST = %f %f %f %f\n", ri.x, ri.y, ri.z, ri.w);
      rd[0] = ri.x;
      rd[1] = ri.y;
      rd[2] = ri.z;
      rd[3] = ri.w;
   }
}

extern void execute_shader(float4 *dests, float4 *inputs,
                           float4 *consts);

void run_vertex_shader(float (*ainputs)[16][4],
                       float (*dests)[16][4],
                       float (*aconsts)[4],
                       int count,
                       int num_attribs)
{
   float4  inputs[16*32*4][16];
   float4  consts[16];
   float4  results[16*32*4][16];

   printf("XXXXXXXXXXX run_vertex_shader\n");
   from_array(inputs, ainputs, count, num_attribs);
   from_consts(consts, aconsts, 5);
   for (int i = 0; i < count; ++i) {
      float4 *in  = inputs[i];
      float4 *res = results[i];
      execute_shader(res, in, consts);
      to_array(dests[i], res, num_attribs);
   }
}