matrix_cubesphere_specular: incomplete

drm/matrix_cubesphere_specular.fs.asm

@@ -0,0 +1,102 @@
+-- temp[0] -- position (world space)
+-- temp[1] -- normal
+-- temp[2] -- light pos (world space)
+-- temp[3] -- view pos (world space)
+-- temp[4] -- texture
+
+-- PIXSIZE 4
+
+TEX TEX_SEM_WAIT TEX_SEM_ACQUIRE
+  temp[4].rgba = LD tex[0].rgba temp[4].rgaa ;
+
+-- normal = normalize(normal)
+-- normal = (1.0 / sqrt(dot(normal, normal))) * normal
+src0.rgb = temp[1] :
+              DP3 src0.rgb src0.rgb ,
+  temp[1].a = DP ;
+src0.a = temp[1] :
+  temp[1].a   = RSQ |src0.a| ;
+src0.a = temp[1], src0.rgb = temp[1] :
+  temp[1].rgb = MAD src0.rgb src0.aaa src0.000 ;
+
+-- light_dir = normalize((f_light_pos - f_world_pos))
+src1.rgb = temp[2] , -- f_light_pos
+src0.rgb = temp[0] , -- f_world_pos
+srcp.rgb = neg :     -- (f_light_pos - f_world_pos)
+              DP3 srcp.rgb srcp.rgb ,
+  temp[2].a = DP ;
+src0.a = temp[2] :
+  temp[2].a   = RSQ |src0.a| ;
+src0.a = temp[2], src0.rgb = temp[2] :
+  temp[2].rgb = MAD src0.rgb src0.aaa src0.000 ;
+
+-- diff = dot(normal, light_dir)
+src0.rgb = temp[2] ,
+src1.rgb = temp[1] :
+                DP3 src0.rgb src1.rgb ,
+   temp[5].a  = DP ;
+
+-- diff = max(diff, 0)
+src0.a = temp[5] :
+  temp[5].a = MAX src0.a src0.0 ;
+
+-- intensity = diff + 0.125
+src0.a = temp[5] ,
+src1.a = float(32) : -- 0.125
+  temp[5].a = MAD src0.a src0.1 src1.a ;
+
+--
+-- specular
+--
+-- temp[3] -- view pos (world space)
+-- view_dir = normalize(f_view_pos - f_world_pos)
+src1.rgb = temp[3] , -- f_light_pos
+src0.rgb = temp[0] , -- f_world_pos
+srcp.rgb = neg :     -- (f_light_pos - f_world_pos)
+              DP3 srcp.rgb srcp.rgb ,
+  temp[3].a = DP ;
+src0.a = temp[3] :
+  temp[3].a   = RSQ |src0.a| ;
+src0.a = temp[3], src0.rgb = temp[3] :
+  temp[3].rgb = MAD src0.rgb src0.aaa src0.000 ;
+
+-- reflect(I, N)
+-- I - 2.0 * dot(N, I) * N
+-- reflect_dir = reflect(-light_dir, norm)
+-- reflect_dir = reflect(-temp[2], temp[1])
+-- I - 2.0 * dot(N, I) * N
+-- - (2.0 * dot(N, I)) * temp[1] + -temp[2]
+src0.rgb = temp[1] , -- N=normal
+src1.rgb = temp[2] : -- I=light_dir   dot(N, -I)
+  temp[6].r = DP3 src0.rgb -src1.rgb ;
+src0.rgb = temp[6] ,
+src1.rgb = float(64) : -- 2.0
+  temp[6].r = MAD src0.r00 src1.r00 src0.000 ;
+src0.rgb = temp[6] ,
+src1.rgb = temp[1] , -- N
+src2.rgb = temp[2] : -- I
+  temp[6].rgb = MAD -src0.rrr src1.rgb -src2.rgb ;
+
+-- spec = max(dot(view_dir, reflect_dir), 0.0)
+src0.rgb = temp[6] , -- reflect_dir
+src1.rgb = temp[3] : -- view_dir
+  temp[6].r = DP3 src0.rgb src1.rgb ;
+src0.rgb = temp[6] :
+  temp[6].a = MAX src0.r src0.0 ;
+
+-- spec = pow(spec, 32)
+src0.a = temp[6] :
+  temp[6].a = LN2 src0.a ;
+src0.a = temp[6] ,
+src1.a = float(72) : -- 32
+  temp[6].a = MAD src0.a src1.a src1.0 ;
+src0.a = temp[6] :
+  temp[6].a = EX2 src0.a ;
+
+OUT TEX_SEM_WAIT
+src0.a = temp[4],
+src0.rgb = temp[4] ,
+src1.a = temp[6] ,
+src1.rgb = temp[6]   :
+  out[0].a    = MAX src0.a src0.a ,
+  out[0].rgb  = MAD src0.111 src1.aaa src1.000 ;

drm/matrix_cubesphere_specular.vs.asm

@@ -0,0 +1,48 @@
+--
+-- dot(m[0], v), dot(m[1], v), dot(m[2], v), dot(m[3], v)
+--
+
+-- input[0] -- position
+-- input[1] -- texture
+-- input[2] -- normal
+
+-- consts[0] -- trans
+-- consts[4] -- world_trans
+-- consts[8] -- normal_trans
+
+-- out[0] -- position clip space
+-- out[1] -- object position world space
+-- out[2] -- normal
+-- out[3] -- light position world space
+-- out[4] -- view position world space
+-- out[5] -- texture
+
+-- position clip space
+temp[1].x   = VE_DOT   const[0].xyzw    input[0].xyzw ;
+temp[1].y   = VE_DOT   const[1].xyzw    input[0].xyzw ;
+temp[1].z   = VE_DOT   const[2].xyzw    input[0].xyzw ;
+temp[1].w   = VE_DOT   const[3].xyzw    input[0].xyzw ;
+
+-- position world space
+temp[2].x   = VE_DOT   const[4].xyzw    input[0].xyzw ;
+temp[2].y   = VE_DOT   const[5].xyzw    input[0].xyzw ;
+temp[2].z   = VE_DOT   const[6].xyzw    input[0].xyzw ;
+temp[2].w   = VE_DOT   const[7].xyzw    input[0].xyzw ;
+
+-- normal world space
+temp[3].x   = VE_DOT   const[4].xyz0    input[2].xyz0 ;
+temp[3].y   = VE_DOT   const[5].xyz0    input[2].xyz0 ;
+temp[3].z   = VE_DOT   const[6].xyz0    input[2].xyz0 ;
+
+-- position (clip space)
+out[0].xyzw = VE_ADD    temp[1].xyzw    const[0].0000 ;
+-- position (world space)
+out[1].xyzw = VE_ADD    temp[2].xyzw    const[0].0000 ;
+-- normal
+out[2].xyzw = VE_ADD    temp[3].xyz0    const[0].0000 ;
+-- light pos (world space)
+out[3].xyzw = VE_ADD   const[8].xyzw    const[8].0000 ;
+-- view pos (world space)
+out[4].xyzw = VE_ADD   const[9].xyzw    const[9].0000 ;
+-- texture
+out[5].xyzw = VE_ADD   input[1].xy00    const[0].0000 ;

regs/us_disassemble2.py

@@ -174,23 +174,24 @@ def disassemble_a_swizzle_sel(code):
     return [mod_str(f"src{alu_sel_strs[sel]}.{swiz}", mod)
             for swiz, sel, mod in zip(a_swiz, a_sels, a_mods)], a_sels
 
-def omod_str(s, mod):
-    if s == 0: # * 1
-        return s
-    elif s == 1: # * 1
-        return f"({s}) * 1"
-    elif s == 2: # * 2
-        return f"({s}) * 2"
-    elif s == 3: # * 4
-        return f"({s}) * 4"
-    elif s == 4: # * 8
-        return f"({s}) * 8"
-    elif s == 5: # / 2
-        return f"({s}) / 2"
-    elif s == 6: # / 4
-        return f"({s}) / 4"
-    elif s == 7: # DISABLE OMOD
-        return s
+def omod_str(mod):
+    if mod == 0: # * 1
+        #return f"1.0 * "
+        return ""
+    elif mod == 1: # * 2
+        return f"2.0 * "
+    elif mod == 2: # * 4
+        return f"4.0 * "
+    elif mod == 3: # * 8
+        return f"8.0 * "
+    elif mod == 4: # / 2
+        return f"0.5 * "
+    elif mod == 5: # / 4
+        return f"0.25 * "
+    elif mod == 6:
+        return f"0.125 * "
+    elif mod == 7: # DISABLE OMOD
+        return "(DISABLE OMOD) "
 
 def disassemble_alu_dest(code):
     a_addrd = US_ALU_ALPHA_INST.ALPHA_ADDRD(code)
@@ -250,14 +251,9 @@ def assert_zeros_common(code):
     assert stat_we == 0
 
 def assert_zeros_alu(code):
-    rgb_omod = US_ALU_RGB_INST.OMOD(code)
     alu_wmask = US_ALU_RGB_INST.ALU_WMASK(code)
-    assert rgb_omod in {0, 7}
     assert alu_wmask == 0
-
-    a_omod = US_ALU_ALPHA_INST.OMOD(code)
     w_omask = US_ALU_ALPHA_INST.W_OMASK(code)
-    assert a_omod in {0, 7}
     assert w_omask == 0
 
 def assert_zeros_tex(code):
@@ -325,6 +321,7 @@ _a_op_operands = {
     "OP_CMP": 3,
     "OP_FRC": 1,
     "OP_EX2": 1,
+    "OP_LN2": 1,
     "OP_RCP": 1,
     "OP_RSQ": 1,
     "OP_SIN": 1,
@@ -387,12 +384,17 @@ def disassemble_alu(code, is_output):
     rgb_clamp_str = ".SAT" if rgb_clamp != 0 else ""
     a_clamp_str = ".SAT" if alpha_clamp != 0 else ""
 
+    rgb_omod = US_ALU_RGB_INST.OMOD(code)
+    a_omod = US_ALU_ALPHA_INST.OMOD(code)
+    rgb_omod_str = omod_str(rgb_omod)
+    a_omod_str = omod_str(a_omod)
+
     print(", ".join([*a_addr_strs, *rgb_addr_strs]), ":")
     #print(", ".join(a_addr_strs), ":")
-    print(f"  {a_out_str}{a_temp_str}{a_op.removeprefix('OP_').ljust(3)}{a_clamp_str} {' '.join(a_swizzle_sel)}", ",")
+    print(f"  {a_out_str}{a_temp_str}{a_omod_str}{a_op.removeprefix('OP_').ljust(3)}{a_clamp_str} {' '.join(a_swizzle_sel)}", ",")
 
     #print(", ".join(rgb_addr_strs), ":")
-    print(f"  {rgb_out_str}{rgb_temp_str}{rgb_op.removeprefix('OP_').ljust(3)}{rgb_clamp_str} {' '.join(rgb_swizzle_sel)}", ";")
+    print(f"  {rgb_out_str}{rgb_temp_str}{rgb_omod_str}{rgb_op.removeprefix('OP_').ljust(3)}{rgb_clamp_str} {' '.join(rgb_swizzle_sel)}", ";")
 
 def disassemble_tex_swizzle_str(code):
     tex_swiz_strs = ["r", "g", "b", "a"]