*/
#define DISASSEM 0
-/* Should be tunable by the driver - do we want to do matrix
- * multiplications with DP4's or with MUL/MAD's? SSE works better
- * with the latter, drivers may differ.
- */
-#define PREFER_DP4 0
-
/* Use uregs to represent registers internally, translate to Mesa's
* expected formats on emit.
const struct state_key *state;
struct gl_vertex_program *program;
GLint max_inst; /** number of instructions allocated for program */
+ GLboolean mvp_with_dp4;
GLuint temp_in_use;
GLuint temp_reserved;
};
-static const struct ureg undef = {
+static const struct ureg undef = {
PROGRAM_UNDEFINED,
0,
0,
{
reg.negate ^= 1;
return reg;
-}
+}
static struct ureg swizzle( struct ureg reg, int x, int y, int z, int w )
GET_SWZ(reg.swz, y),
GET_SWZ(reg.swz, z),
GET_SWZ(reg.swz, w));
-
return reg;
}
src->File = reg.file;
src->Index = reg.idx;
src->Swizzle = reg.swz;
- src->NegateBase = reg.negate ? NEGATE_XYZW : 0;
+ src->Negate = reg.negate ? NEGATE_XYZW : NEGATE_NONE;
src->Abs = 0;
- src->NegateAbs = 0;
src->RelAddr = 0;
/* Check that bitfield sizes aren't exceeded */
ASSERT(src->Index == reg.idx);
inst = &p->program->Base.Instructions[nr];
inst->Opcode = (enum prog_opcode) op;
- inst->StringPos = 0;
inst->Data = 0;
emit_arg( &inst->SrcReg[0], src0 );
p->eye_position = reserve_temp(p);
- if (PREFER_DP4) {
+ if (p->mvp_with_dp4) {
register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
0, modelview );
struct ureg hpos = register_output( p, VERT_RESULT_HPOS );
struct ureg mvp[4];
- if (PREFER_DP4) {
+ if (p->mvp_with_dp4) {
register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3,
0, mvp );
emit_matrix_transform_vec4( p, hpos, mvp, pos );
static GLuint material_attrib( GLuint side, GLuint property )
{
- return ((property - STATE_AMBIENT) * 2 +
- side);
+ return (property - STATE_AMBIENT) * 2 + side;
}
struct ureg material_ambient = get_material(p, side, STATE_AMBIENT);
struct ureg material_diffuse = get_material(p, side, STATE_DIFFUSE);
struct ureg tmp = make_temp(p, material_diffuse);
- emit_op3(p, OPCODE_MAD, tmp, WRITEMASK_XYZ, lm_ambient,
+ emit_op3(p, OPCODE_MAD, tmp, WRITEMASK_XYZ, lm_ambient,
material_ambient, material_emission);
return tmp;
}
register_param3(p, STATE_LIGHT, light, property);
struct ureg material_value = get_material(p, side, property);
struct ureg tmp = get_temp(p);
- emit_op2(p, OPCODE_MUL, tmp, 0, light_value, material_value);
+ emit_op2(p, OPCODE_MUL, tmp, 0, light_value, material_value);
return tmp;
}
else
/* Calculate distance attenuation:
*/
if (p->state->unit[i].light_attenuated) {
-
/* 1/d,d,d,1/d */
emit_op1(p, OPCODE_RCP, dist, WRITEMASK_YZ, dist);
/* 1,d,d*d,1/d */
emit_op1(p, OPCODE_RCP, dist, 0, dist);
/* spot-atten * dist-atten */
emit_op2(p, OPCODE_MUL, att, 0, dist, att);
- } else {
+ }
+ else {
/* dist-atten */
emit_op1(p, OPCODE_RCP, att, 0, dist);
}
/*
* NOTE:
- * dot.x = dot(normal, VPpli)
- * dot.y = dot(normal, halfAngle)
- * dot.z = back.shininess
- * dot.w = front.shininess
+ * dots.x = dot(normal, VPpli)
+ * dots.y = dot(normal, halfAngle)
+ * dots.z = back.shininess
+ * dots.w = front.shininess
*/
for (i = 0; i < MAX_LIGHTS; i++)
{
if (!p->state->material_shininess_is_zero) {
struct ureg shininess = get_material(p, 0, STATE_SHININESS);
- emit_op1(p, OPCODE_MOV, dots, WRITEMASK_W, swizzle1(shininess,X));
+ emit_op1(p, OPCODE_MOV, dots, WRITEMASK_W, swizzle1(shininess,X));
release_temp(p, shininess);
}
_col1 = make_temp(p, get_identity_param(p));
else
_col1 = _col0;
-
}
if (twoside) {
if (!p->state->material_shininess_is_zero) {
+ /* Note that we negate the back-face specular exponent here.
+ * The negation will be un-done later in the back-face code below.
+ */
struct ureg shininess = get_material(p, 1, STATE_SHININESS);
emit_op1(p, OPCODE_MOV, dots, WRITEMASK_Z,
negate(swizzle1(shininess,X)));
half = get_temp(p);
emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
emit_normalize_vec3(p, half, half);
- } else {
+ }
+ else {
half = register_param3(p, STATE_INTERNAL,
STATE_LIGHT_HALF_VECTOR, i);
}
}
- }
+ }
else {
struct ureg Ppli = register_param3(p, STATE_INTERNAL,
STATE_LIGHT_POSITION, i);
res0 = _col0;
res1 = register_output( p, VERT_RESULT_COL0 );
}
- } else {
+ }
+ else {
mask0 = 0;
mask1 = 0;
res0 = _col0;
emit_op1(p, OPCODE_LIT, lit, 0, dots);
emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
emit_op3(p, OPCODE_MAD, _col0, 0, swizzle1(lit,X), ambient, _col0);
- }
+ }
else if (!p->state->material_shininess_is_zero) {
/* there's a non-zero specular term */
emit_op1(p, OPCODE_LIT, lit, 0, dots);
emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
- }
+ }
else {
/* no attenutation, no specular */
emit_degenerate_lit(p, lit, dots);
res0 = _bfc0;
res1 = register_output( p, VERT_RESULT_BFC0 );
}
- } else {
+ }
+ else {
res0 = _bfc0;
res1 = _bfc1;
mask0 = 0;
mask1 = 0;
}
+ /* For the back face we need to negate the X and Y component
+ * dot products. dots.Z has the negated back-face specular
+ * exponent. We swizzle that into the W position. This
+ * negation makes the back-face specular term positive again.
+ */
dots = negate(swizzle(dots,X,Y,W,Z));
if (!is_undef(att)) {
}
else if (!p->state->material_shininess_is_zero) {
emit_op1(p, OPCODE_LIT, lit, 0, dots);
- emit_op2(p, OPCODE_ADD, _bfc0, 0, ambient, _bfc0);
- }
+ emit_op2(p, OPCODE_ADD, _bfc0, 0, ambient, _bfc0); /**/
+ }
else {
emit_degenerate_lit(p, lit, dots);
emit_op2(p, OPCODE_ADD, _bfc0, 0, ambient, _bfc0);
emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _bfc0);
emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _bfc1);
- /* restore negate flag for next lighting */
- dots = negate(dots);
+ /* restore dots to its original state for subsequent lights
+ * by negating and swizzling again.
+ */
+ dots = negate(swizzle(dots,X,Y,W,Z));
release_temp(p, ambient);
release_temp(p, diffuse);
struct ureg in = (!is_undef(out_texgen) ?
out_texgen :
register_input(p, VERT_ATTRIB_TEX0+i));
- if (PREFER_DP4) {
+ if (p->mvp_with_dp4) {
register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
0, texmat );
emit_matrix_transform_vec4( p, out, texmat, in );
}
release_temps(p);
- }
+ }
else {
emit_passthrough(p, VERT_ATTRIB_TEX0+i, VERT_RESULT_TEX0+i);
}
static void build_tnl_program( struct tnl_program *p )
-{ /* Emit the program, starting with modelviewproject:
+{
+ /* Emit the program, starting with modelviewproject:
*/
build_hpos(p);
static void
create_new_program( const struct state_key *key,
struct gl_vertex_program *program,
+ GLboolean mvp_with_dp4,
GLuint max_temps)
{
struct tnl_program p;
p.transformed_normal = undef;
p.identity = undef;
p.temp_in_use = 0;
+ p.mvp_with_dp4 = mvp_with_dp4;
if (max_temps >= sizeof(int) * 8)
p.temp_reserved = 0;
return NULL;
create_new_program( &key, prog,
+ ctx->mvp_with_dp4,
ctx->Const.VertexProgram.MaxTemps );
#if 0
projects / mesa.git / blobdiff