Saving state.

src/liberty/collision/ColLine.h

@@ -1,14 +1,14 @@
 #pragma once
 
-struct CColLine
+struct alignas(8) CColLine
 {
 	// NB: this has to be compatible with two CVuVectors
-	CVector p0;
+	alignas(8) CVector p0;
-	int pad0;
+	float pad0 = 0.0f;
-	CVector p1;
+	alignas(8) CVector p1;
-	int pad1;
+	float pad1 = 0.0f;
 
-	CColLine(void) { };
+	CColLine(void) = default;
 	CColLine(const CVector &p0, const CVector &p1) { this->p0 = p0; this->p1 = p1; };
 	void Set(const CVector &p0, const CVector &p1);
 };

src/liberty/collision/ColPoint.h

@@ -1,12 +1,12 @@
 #pragma once
 
-struct CColPoint
+struct alignas(8) CColPoint
 {
-	CVector point;
+	alignas(8) CVector point;
-	int pad1;
+	float pad1 = 0.0f;
 	// the surface normal on the surface of point
-	CVector normal;
+	alignas(8) CVector normal;
-	int pad2;
+	float pad2 = 0.0f;
 	uint8 surfaceA;
 	uint8 pieceA;
 	uint8 surfaceB;

src/liberty/collision/Collision.cpp

@@ -787,12 +787,13 @@ bool
 CCollision::ProcessSphereSphere(const CColSphere &s1, const CColSphere &s2, CColPoint &point, float &mindistsq)
 {
 	CVector dist = s1.center - s2.center;
-	float d = dist.Magnitude() - s2.radius;	// distance from s1's center to s2
+	float mag = dist.Magnitude();
+	float d = mag - s2.radius;	// distance from s1's center to s2
 	float depth = s1.radius - d;	// sphere overlap
 	if(d < 0.0f) d = 0.0f;		// clamp to zero, i.e. if s1's center is inside s2
 	// no collision if sphere is not close enough
 	if(d*d < mindistsq && d < s1.radius){
-		dist.Normalise();
+		dist *= Invert<true, false>(mag);
 		point.point = s1.center - dist*d;
 		point.normal = dist;
 #ifndef VU_COLLISION

src/liberty/math/Quaternion.cpp

@@ -39,10 +39,9 @@ CQuaternion::Slerp(const CQuaternion &q1, const CQuaternion &q2, float theta, fl
 void
 CQuaternion::Multiply(const CQuaternion &q1, const CQuaternion &q2)
 {
-	x = (q2.z * q1.y) - (q1.z * q2.y) + (q1.x * q2.w) + (q2.x * q1.w);
+	quat_mult(reinterpret_cast<quaternion_t *>(this), 
-	y = (q2.x * q1.z) - (q1.x * q2.z) + (q1.y * q2.w) + (q2.y * q1.w);
+			  reinterpret_cast<const quaternion_t &>(q1),
-	z = (q2.y * q1.x) - (q1.y * q2.x) + (q1.z * q2.w) + (q2.z * q1.w);
+			  reinterpret_cast<const quaternion_t &>(q2));
-	w = (q2.w * q1.w) - (q2.x * q1.x) - (q2.y * q1.y) - (q2.z * q1.z);
 }
 
 void
@@ -50,10 +49,11 @@ CQuaternion::Get(RwV3d *axis, float *angle)
 {
 	*angle = Acos(w);
 	float s = Sin(*angle);
+	float invS = dc::Invert<true, false>(s);
 
-	axis->x = x * (1.0f / s);
+	axis->x = x * invS;
-	axis->y = y * (1.0f / s);
+	axis->y = y * invS;
-	axis->z = z * (1.0f / s);
+	axis->z = z * invS;
 }
 
 void
@@ -104,7 +104,7 @@ CQuaternion::Set(const RwMatrix &matrix)
 	if (f >= 0.0f) {
 		s = Sqrt(f + 1.0f);
 		w = 0.5f * s;
-		m = 0.5f / s;
+		m = Div<true, false>(0.5f, s);
 		x = (matrix.up.z - matrix.at.y) * m;
 		y = (matrix.at.x - matrix.right.z) * m;
 		z = (matrix.right.y - matrix.up.x) * m;
@@ -115,7 +115,7 @@ CQuaternion::Set(const RwMatrix &matrix)
 	if (f >= 0.0f) {
 		s = Sqrt(f + 1.0f);
 		x = 0.5f * s;
-		m = 0.5f / s;
+		m = Div<true, false>(0.5f, s);
 		y = (matrix.up.x + matrix.right.y) * m;
 		z = (matrix.at.x + matrix.right.z) * m;
 		w = (matrix.up.z - matrix.at.y) * m;
@@ -126,7 +126,7 @@ CQuaternion::Set(const RwMatrix &matrix)
 	if (f >= 0.0f) {
 		s = Sqrt(f + 1.0f);
 		y = 0.5f * s;
-		m = 0.5f / s;
+		m = Div<true, false>(0.5f, s);
 		w = (matrix.at.x - matrix.right.z) * m;
 		x = (matrix.up.x - matrix.right.y) * m;
 		z = (matrix.at.y + matrix.up.z) * m;
@@ -136,7 +136,7 @@ CQuaternion::Set(const RwMatrix &matrix)
 	f = matrix.at.z - (matrix.up.y + matrix.right.x);
 	s = Sqrt(f + 1.0f);
 	z = 0.5f * s;
-	m = 0.5f / s;
+	m = Div<true, false>(0.5f, s);
 	w = (matrix.right.y - matrix.up.x) * m;
 	x = (matrix.at.x + matrix.right.z) * m;
 	y = (matrix.at.y + matrix.up.z) * m;
@@ -151,8 +151,7 @@ CQuaternion::Get(float *f1, float *f2, float *f3)
 	*f3 = Atan2(matrix.right.y, matrix.up.y);
 	if (*f3 < 0.0f)
 		*f3 += TWOPI;
-	float s = Sin(*f3);
+	auto [s, c] = SinCos(*f3);
-	float c = Cos(*f3);
 	*f1 = Atan2(-matrix.at.y, s * matrix.right.y + c * matrix.up.y);
 	if (*f1 < 0.0f)
 		*f1 += TWOPI;
@@ -164,12 +163,10 @@ CQuaternion::Get(float *f1, float *f2, float *f3)
 void
 CQuaternion::Set(float f1, float f2, float f3)
 {
-	float c1 = Cos(f1 * 0.5f);
+	auto [s1, c1] = SinCos(f1 * 0.5f);
-	float c2 = Cos(f2 * 0.5f);
+	auto [s2, c2] = SinCos(f2 * 0.5f);
-	float c3 = Cos(f3 * 0.5f);
+	auto [s3, c3] = SinCos(f3 * 0.5f);
-	float s1 = Sin(f1 * 0.5f);
+
-	float s2 = Sin(f2 * 0.5f);
-	float s3 = Sin(f3 * 0.5f);
 	x = ((c2 * c1) * s3) - ((s2 * s1) * c3);
 	y = ((s1 * c2) * c3) + ((s2 * c1) * s3);
 	z = ((s2 * c1) * c3) - ((s1 * c2) * s3);

src/miami/collision/Collision.cpp

@@ -617,12 +617,13 @@ bool
 CCollision::ProcessSphereSphere(const CColSphere &s1, const CColSphere &s2, CColPoint &point, float &mindistsq)
 {
 	CVector dist = s1.center - s2.center;
-	float d = dist.Magnitude() - s2.radius;	// distance from s1's center to s2
+	float mag = dist.Magnitude();
+	float d = mag - s2.radius;	// distance from s1's center to s2
 	float depth = s1.radius - d;	// sphere overlap
 	if(d < 0.0f) d = 0.0f;		// clamp to zero, i.e. if s1's center is inside s2
 	// no collision if sphere is not close enough
 	if(d*d < mindistsq && d < s1.radius){
-		dist.Normalise();
+		dist *= Invert<true, false>(mag);
 		point.point = s1.center - dist*d;
 		point.normal = dist;
 #ifndef VU_COLLISION

src/miami/math/Quaternion.cpp

@@ -49,7 +49,7 @@ CQuaternion::Get(RwV3d *axis, float *angle)
 {
 	*angle = Acos(w);
 	float s = Sin(*angle);
-	float invS = dc::Invert(s);
+	float invS = dc::Invert<true, false>(s);
 
 	axis->x = x * invS;
 	axis->y = y * invS;
@@ -151,8 +151,7 @@ CQuaternion::Get(float *f1, float *f2, float *f3)
 	*f3 = Atan2(matrix.right.y, matrix.up.y);
 	if (*f3 < 0.0f)
 		*f3 += TWOPI;
-	float s = Sin(*f3);
+	auto [s, c] = SinCos(*f3);
-	float c = Cos(*f3);
 	*f1 = Atan2(-matrix.at.y, s * matrix.right.y + c * matrix.up.y);
 	if (*f1 < 0.0f)
 		*f1 += TWOPI;
@@ -164,12 +163,10 @@ CQuaternion::Get(float *f1, float *f2, float *f3)
 void
 CQuaternion::Set(float f1, float f2, float f3)
 {
-	float c1 = Cos(f1 * 0.5f);
+	auto [s1, c1] = SinCos(f1 * 0.5f);
-	float c2 = Cos(f2 * 0.5f);
+	auto [s2, c2] = SinCos(f2 * 0.5f);
-	float c3 = Cos(f3 * 0.5f);
+	auto [s3, c3] = SinCos(f3 * 0.5f);
-	float s1 = Sin(f1 * 0.5f);
+
-	float s2 = Sin(f2 * 0.5f);
-	float s3 = Sin(f3 * 0.5f);
 	x = ((c2 * c1) * s3) - ((s2 * s1) * c3);
 	y = ((s1 * c2) * c3) + ((s2 * c1) * s3);
 	z = ((s2 * c1) * c3) - ((s1 * c2) * s3);

vendor/dca3-kos

@@ -1 +1 @@
-Subproject commit 5d475f6ecffa0bc7aaaccf9bd8b64cda09aec89d
+Subproject commit 5c0e0936ad15a249858b01f29210a2ae7beec603

vendor/librw/src/dc/rwdc.cpp

@@ -2856,7 +2856,7 @@ void tnlMeshletSkinVertices(uint8_t *OCR, uint8_t *OCR_normal, const uint8_t* ve
 
 				__builtin_prefetch(srcVtxBytes);
 				for(int c = 0; c < count - 1; ++c)
-					innerLoop.template operator()<false>();
+					innerLoop.template operator()<true>();
 				innerLoop.template operator()<false>();
 	
 			} else if (!(flags & 0x80)) {
@@ -3464,9 +3464,6 @@ void defaultRenderCB(ObjPipeline *pipe, Atomic *atomic) {
 	assert(atomicContexts.size() <= 32767);
 	auto meshes = geo->meshHeader->getMeshes();
 
-	RawMatrix worldOrient;
-	bool worldOrientValid = false;
-
 	for (int16_t n = 0; n < numMeshes; n++) {
 		bool doBlend = meshes[n].material->color.alpha != 255; // TODO: check all vertexes for alpha?
 		bool doBlendMaterial = doBlend;
@@ -3485,16 +3482,8 @@ void defaultRenderCB(ObjPipeline *pipe, Atomic *atomic) {
 			isMatFX = true;
 			matfxCoefficient = matfx->fx[0].env.coefficient;
 			matfxContexts.resize(matfxContexts.size() + 1);
-#warning "Get rid of me by creating mat_apply_3x3()!"
+
-			if(!worldOrientValid) {
+			uploadEnvMatrix(matfx->fx[0].env.frame, reinterpret_cast<rw::RawMatrix*>(atomic->getFrame()->getLTM()), &matfxContexts.back().mtx);
-				rw::convMatrix(&worldOrient, atomic->getFrame()->getLTM());
-				worldOrient.pos = { 0, 0, 0 };
-				worldOrient.rightw = 0;
-				worldOrient.upw = 0;
-				worldOrient.atw = 0;
-				worldOrientValid = true;
-			}
-			uploadEnvMatrix(matfx->fx[0].env.frame, &worldOrient, &matfxContexts.back().mtx);
 			matfxContexts.back().coefficient = matfxCoefficient;
 			
 			pvr_poly_cxt_t cxt;

vendor/librw/src/dc/rwdc_common.h

@@ -267,62 +267,6 @@ __always_inline __hot constexpr float Acos(float x) {
         x2 = __x; y2 = __y; z2 = __z; \
     }
 
-// no declspec naked, so can't do rts / fschg. instead compiler pads with nop?
-__always_inline __hot void mat_load_3x3(const matrix_t* mtx) {
-    __asm__ __volatile__ (
-        R"(
-            fschg
-            frchg
-
-            fmov        @%[mtx]+, dr0
-            fldi0 		fr12
-
-            fmov        @%[mtx]+, dr2
-            fldi0 		fr13
-
-            fmov        @%[mtx]+, dr4
-            fldi0	    fr3
-
-            fmov        @%[mtx]+, dr6
-            fmov        dr12, dr14
-
-            fmov        @%[mtx]+, dr8
-            fldi0	    fr7
-
-            fmov        @%[mtx]+, dr10
-            fldi0	    fr11
-
-            fschg
-            frchg
-        )"
-        : [mtx] "+r" (mtx)
-    );
-}
-
-// sets pos.w to 1
-__always_inline __hot void rw_mat_load_4x4(const rw::Matrix* mtx) {
-    __asm__ __volatile__ (
-        R"(
-            fschg
-            frchg
-            fmov        @%[mtx]+, dr0
-
-            fmov        @%[mtx]+, dr2
-            fmov        @%[mtx]+, dr4
-            fmov        @%[mtx]+, dr6
-            fmov        @%[mtx]+, dr8
-            fmov        @%[mtx]+, dr10
-            fmov        @%[mtx]+, dr12
-            fmov        @%[mtx]+, dr14
-            fldi1 	 	fr15
-
-            fschg
-            frchg
-        )"
-        : [mtx] "+r" (mtx)
-    );
-}
-
 __always_inline __hot void mat_transpose(void) {
     asm volatile (
         "frchg\n\t" // fmov for singles only works on front bank
@@ -560,23 +504,6 @@ __hot inline void mat_load_apply(const matrix_t* matrix1, const matrix_t* matrix
 		w_ = tmp1233123.w; \
 	} while(false)
 
-inline void mat_load_3x3(const matrix_t* mtx) {
-	memcpy(XMTRX, mtx, sizeof(matrix_t));
-	XMTRX[0][3] = 0.0f;
-	XMTRX[1][3] = 0.0f;
-	XMTRX[2][3] = 0.0f;
-
-	XMTRX[3][0] = 0.0f;
-	XMTRX[3][1] = 0.0f;
-	XMTRX[3][2] = 0.0f;
-	XMTRX[3][3] = 0.0f;
-}
-
-inline void rw_mat_load_4x4(const rw::Matrix* mtx) {
-	memcpy(XMTRX, mtx, sizeof(matrix_t));
-	XMTRX[3][3] = 1.0f;
-}
-
 inline void mat_transpose(void) {
     matrix_t tmp;