wip

2025-11-11 08:27:15 -06:00 · 2025-11-11 08:27:15 -06:00 · db0574345b
commit db0574345b
parent 399cd6aaf9
3 changed files with 252 additions and 8 deletions
--- a/drm/shadertoy.c
+++ b/drm/shadertoy.c
@ -409,7 +409,7 @@ int indirect_buffer(float time)
  };
  const int vertex_shader_length = (sizeof (vertex_shader)) / (sizeof (vertex_shader[0]));
  assert(vertex_shader_length % 4 == 0);
-  printf("vs length %d\n", vertex_shader_length);
+  //printf("vs length %d\n", vertex_shader_length);

  T0_ONE_REG(VAP_PVS_VECTOR_DATA_REG_128, vertex_shader_length - 1);
  for (int i = 0; i < vertex_shader_length; i++) {
@ -449,10 +449,17 @@ int indirect_buffer(float time)

  // fragment constants

+#define PI (3.14159274101257324219f)
+#define PI_2 (PI * 2.0f)
+#define I_PI_2 (1.0f / (PI_2))
+
  const float fragment_consts[] = {
-    time, 0, 0, 0,
+    time, 1.2, 0.01, 0.4,
+    PI_2, I_PI_2, 0, 0,
+    0.25, 0.40625, 0.5625, 0,
  };
  int fragment_consts_length = (sizeof (fragment_consts)) / (sizeof (fragment_consts[0]));
+  assert(fragment_consts_length % 4 == 0);

  T0V(GA_US_VECTOR_INDEX
      , GA_US_VECTOR_INDEX__INDEX(0)
@ -465,16 +472,16 @@ int indirect_buffer(float time)
  // fragment code

  const uint32_t fragment_shader[] = {
-    #include "shadertoy_palette.fs.inc"
+    #include "shadertoy_palette_fractal.fs.inc"
  };
  const int fragment_shader_length = (sizeof (fragment_shader)) / (sizeof (fragment_shader[0]));
  assert(fragment_shader_length % 6 == 0);
-  printf("fs length %d\n", fragment_shader_length);
+  //printf("fs length %d\n", fragment_shader_length);
  const int fragment_shader_instructions = fragment_shader_length / 6;
-  printf("fs instructions %d\n", fragment_shader_instructions);
+  //printf("fs instructions %d\n", fragment_shader_instructions);

  T0V(US_PIXSIZE
-      , US_PIXSIZE__PIX_SIZE(2) // pixel shader stack frame size
+      , US_PIXSIZE__PIX_SIZE(3) // pixel shader stack frame size
      );

  T0V(US_CODE_RANGE
@ -513,7 +520,7 @@ int indirect_buffer(float time)
    -1.0f,  1.0f, 0.0f
  };
  const int vertices_length = (sizeof (vertices)) / (sizeof (vertices[0]));
-  printf("vtx length %d\n", vertices_length);
+  //printf("vtx length %d\n", vertices_length);
  T3(_3D_DRAW_IMMD_2, (1 + vertices_length) - 1);
  ib[ix++].u32
    = VAP_VF_CNTL__PRIM_TYPE(4)
@ -708,7 +715,7 @@ int main()
 #define D1GRPH_UPDATE__D1GRPH_SURFACE_UPDATE_PENDING (1 << 2)

    uint32_t d1crtc_double_buffer_control = rreg(rmmio, D1CRTC_DOUBLE_BUFFER_CONTROL);
-    printf("D1CRTC_DOUBLE_BUFFER_CONTROL: %08x\n", d1crtc_double_buffer_control);
+    //printf("D1CRTC_DOUBLE_BUFFER_CONTROL: %08x\n", d1crtc_double_buffer_control);
    assert(d1crtc_double_buffer_control == (1 << 8));

    // addresses were retrieved from /sys/kernel/debug/radeon_vram_mm
--- a/drm/shadertoy_palette_fractal.fs.asm
+++ b/drm/shadertoy_palette_fractal.fs.asm
@ -0,0 +1,104 @@
+-- CONST[0] = { time, 1.2, 0.01, 0.4 }
+-- CONST[1] = { PI_2, I_PI_2, 0, 0 },
+-- CONST[2] = { 0.25, 0.40625, 0.5625, 0 },
+
+-- temp[0] : { uv0.xy , _, l }
+-- temp[1] : { uv.xy  , _, d }
+-- temp[2] : final_color.xyzw
+-- temp[3] : {col.xyz , i }
+
+-- vec2 uv = uv0; // temp[1]
+src0.rgb = temp[0] : -- uv0
+  temp[1].rg  = MAX src0.rg_ src0.rg_ ;
+
+-- vec4 final_color = vec4(0, 0, 0, 1);
+:
+  temp[2].a   = MAX src0.1 src0.1 ,
+  temp[2].rgb = MAX src0.000 src0.000 ;
+
+-- i = 0;
+:
+  temp[3].a   = MAX src0.0 src0.0 ;
+
+--------------------------------------------------------------------------------
+-- loop start
+--------------------------------------------------------------------------------
+
+-- uv = uv * 1.5;
+src0.rgb = temp[1] , -- uv
+src0.a = float(60) : -- 1.5
+  temp[1].rg  = MAD src0.rg_ src0.aa_ src0.00_ ;
+
+-- uv = fract(uv);
+src0.rgb = temp[1] : -- uv
+  temp[1].rg  = FRC src0.rg_ ;
+
+-- uv = uv - 0.5;
+src0.a = float(48) : -- 0.5
+  temp[1].rg  = MAD src0.rg_ src0.11_ src0.aa_ ;
+
+-- l = length(uv0);
+src0.rgb = temp[0] : -- uv0
+                 DP3 src0.rg0 src0.rg0 ,
+  temp[0].a    = DP ;
+src0.a   = temp[0] :
+  temp[0].a    = RSQ |src0.a| ;
+src0.a   = temp[0] :
+  temp[0].a    = RCP src0.a ;
+
+-- d = i * 0.4 + l;
+src0.a = const[0] , -- 0.4
+src1.a = temp[0]  , -- l
+src2.a = temp[3]  : -- i
+  temp[1].a   = MAD src2.a src0.a src1.a ;
+
+-- d = time * 0.4 + d;
+src0.a = const[0]   , -- 0.4
+src1.a = temp[1]    , -- d
+src0.rgb = const[0] : -- time (r)
+  temp[1].a   = MAD src0.r src0.a src1.a ;
+
+-- palette
+-- v = d + (vec3(0.25, 0.40625, 0.5625) + 0.5)
+src0.a = temp[1]     , -- d
+src0.rgb = const[2]  , -- vec3(0.25, 0.40625, 0.5625)
+src1.rgb = float(48) , -- 0.5
+srcp.rgb = add       : -- (vec3(0.25, 0.40625, 0.5625) + 0.5)
+  temp[3].rgb = MAD src0.111 src0.aaa srcp.rgb ;
+
+-- v = frac(v)
+src0.rgb = temp[3] : -- v
+  temp[3].rgb = FRC src0.rgb ;
+
+src0.rgb = temp[3]   , -- v
+src1.rgb = float(48) : -- 0.5
+  temp[3].rgb = MAD src0.111 src0.rgb -src1.rgb ;
+
+-- v = cos(v)
+src0.rgb = temp[3] : -- v
+              COS src0.r ,
+  temp[3].r = SOP ;
+src0.rgb = temp[3] : -- v
+              COS src0.g ,
+  temp[3].g = SOP ;
+src0.rgb = temp[3] : -- v
+              COS src0.b ,
+  temp[3].b = SOP ;
+
+-- col = vec3(0.5, 0.5, 0.5) * v + vec3(0.5, 0.5, 0.5)
+src0.rgb = temp[3]   , -- v
+src1.rgb = float(48) : -- 0.5
+  temp[3].rgb = MAD src1.rgb src0.rgb src1.rgb;
+
+-- end of 'palette' function
+
+-----
+-----
+-----
+
+OUT TEX_SEM_WAIT
+src0.a   = temp[0] ,
+src0.rgb = temp[0] ,
+src1.rgb = temp[3] :
+  out[0].a     = MAX src0.1 src0.1 ,
+  out[0].rgb   = MAX src1.rgb src1.rgb ;
--- a/drm/shadertoy_palette_fractal.fs.inc
+++ b/drm/shadertoy_palette_fractal.fs.inc
@ -0,0 +1,133 @@
+0x00001800,
+0x08020000,
+0x08020080,
+0x00e40720,
+0x00000000,
+0x00000015,
+
+0x00007800,
+0x08020080,
+0x08020080,
+0x00920490,
+0x00c18023,
+0x00000025,
+
+0x00004000,
+0x08020080,
+0x08020080,
+0x00000000,
+0x00810033,
+0x00000000,
+
+0x00001800,
+0x08020001,
+0x080200bc,
+0x00ed8720,
+0x00000000,
+0x00790010,
+
+0x00001800,
+0x08020001,
+0x08020080,
+0x00000720,
+0x00000000,
+0x00000019,
+
+0x00001800,
+0x08020080,
+0x080200b0,
+0x00fb0720,
+0x00000000,
+0x0076c010,
+
+0x00004000,
+0x08020000,
+0x08020080,
+0x00840420,
+0x00000001,
+0x00000001,
+
+0x00004000,
+0x08020080,
+0x08020000,
+0x00000000,
+0x0004c00b,
+0x00000000,
+
+0x00004000,
+0x08020080,
+0x08020000,
+0x00000000,
+0x0000c00a,
+0x00000000,
+
+0x00004000,
+0x08020080,
+0x00300100,
+0x00000000,
+0x0060e010,
+0x1a000000,
+
+0x00004000,
+0x08020100,
+0x08000500,
+0x00000000,
+0x00600010,
+0x1a000000,
+
+0x00003800,
+0x8802c102,
+0x08020001,
+0x006d86d8,
+0x00000000,
+0x00223030,
+
+0x00003800,
+0x08020003,
+0x08020080,
+0x00000220,
+0x00000000,
+0x00000039,
+
+0x00003800,
+0x0802c003,
+0x08020080,
+0x004406d8,
+0x00000000,
+0x00a21030,
+
+0x00000800,
+0x08020003,
+0x08020080,
+0x00000000,
+0x0000000d,
+0x0000003a,
+
+0x00001000,
+0x08020003,
+0x08020080,
+0x00000000,
+0x0000400d,
+0x0000003a,
+
+0x00002000,
+0x08020003,
+0x08020080,
+0x00000000,
+0x0000800d,
+0x0000003a,
+
+0x00003800,
+0x0802c003,
+0x08020080,
+0x00440221,
+0x00000000,
+0x00221030,
+
+0x00078005,
+0x08000c00,
+0x08020000,
+0x00442221,
+0x00c18003,
+0x00000005,
+