Using NEON optimized code where possible.

src/base.cpp

@@ -20,6 +20,9 @@
 #include "rwengine.h"
 
 #include <vitaGL.h>
+extern "C"{
+#include <math_neon.h>
+};
 
 namespace rw {
 
@@ -190,6 +193,9 @@ V3d::transformVectors(V3d *out, const V3d *in, int32 n, const Matrix *m)
 void
 RawMatrix::mult(RawMatrix *dst, RawMatrix *src1, RawMatrix *src2)
 {
+#ifdef PSP2
+	matmul4_neon((float*)&dst->right.x, (float*)&src1->right.x, (float*)&src2->right.x);
+#else
 	dst->right.x = src1->right.x*src2->right.x + src1->right.y*src2->up.x + src1->right.z*src2->at.x + src1->rightw*src2->pos.x;
 	dst->right.y = src1->right.x*src2->right.y + src1->right.y*src2->up.y + src1->right.z*src2->at.y + src1->rightw*src2->pos.y;
 	dst->right.z = src1->right.x*src2->right.z + src1->right.y*src2->up.z + src1->right.z*src2->at.z + src1->rightw*src2->pos.z;
@@ -206,6 +212,7 @@ RawMatrix::mult(RawMatrix *dst, RawMatrix *src1, RawMatrix *src2)
 	dst->pos.y   = src1->pos.x*src2->right.y   + src1->pos.y*src2->up.y   + src1->pos.z*src2->at.y + src1->posw*src2->pos.y;
 	dst->pos.z   = src1->pos.x*src2->right.z   + src1->pos.y*src2->up.z   + src1->pos.z*src2->at.z + src1->posw*src2->pos.z;
 	dst->posw    = src1->pos.x*src2->rightw    + src1->pos.y*src2->upw    + src1->pos.z*src2->atw  + src1->posw*src2->posw;
+#endif
 }
 
 void

src/rwbase.h

@@ -5,7 +5,9 @@
 #ifndef M_PI
 #define M_PI 3.14159265358979323846
 #endif
-
+extern "C"{
+#include <math_neon.h>
+};
 // TODO: clean up the opengl defines
 //       and figure out what we even want here...
 #ifdef RW_GL3
@@ -240,7 +242,7 @@ inline V3d add(const V3d &a, const V3d &b) { return makeV3d(a.x+b.x, a.y+b.y, a.
 inline V3d sub(const V3d &a, const V3d &b) { return makeV3d(a.x-b.x, a.y-b.y, a.z-b.z); }
 inline V3d scale(const V3d &a, float32 r) { return makeV3d(a.x*r, a.y*r, a.z*r); }
 inline float32 length(const V3d &v) { return sqrtf(v.x*v.x + v.y*v.y + v.z*v.z); }
-inline V3d normalize(const V3d &v) { return scale(v, 1.0f/length(v)); }
+inline V3d normalize(const V3d &v) { V3d r; normalize3_neon((float*)&v.x, &r.x); return r; }
 inline V3d setlength(const V3d &v, float32 l) { return scale(v, l/length(v)); }
 V3d cross(const V3d &a, const V3d &b);
 inline float32 dot(const V3d &a, const V3d &b) { return a.x*b.x + a.y*b.y + a.z*b.z; }
@@ -273,7 +275,10 @@ struct Quat
 	float32 x, y, z, w;
 
 	static Quat rotation(float32 angle, const V3d &axis){
-		return makeQuat(cosf(angle/2.0f), scale(normalize(axis), sinf(angle/2.0f))); }
+		float cs[2];
+		sincosf_c(angle/2.0f, cs);
+		return makeQuat(cs[1], scale(normalize(axis), cs[0]));
+		}
 	void set(float32 w, float32 x, float32 y, float32 z){
 		this->w = w; this->x = x; this->y = y; this->z = z; }
 	V3d vec(void){ return makeV3d(x, y, z); }
@@ -286,10 +291,10 @@ inline Quat makeQuat(float32 w, const V3d &vec) { Quat q = { vec.x, vec.y, vec.z
 inline Quat add(const Quat &q, const Quat &p) { return makeQuat(q.w+p.w, q.x+p.x, q.y+p.y, q.z+p.z); }
 inline Quat sub(const Quat &q, const Quat &p) { return makeQuat(q.w-p.w, q.x-p.x, q.y-p.y, q.z-p.z); }
 inline Quat negate(const Quat &q) { return makeQuat(-q.w, -q.x, -q.y, -q.z); }
-inline float32 dot(const Quat &q, const Quat &p) { return q.w*p.w + q.x*p.x + q.y*p.y + q.z*p.z; }
+inline float32 dot(const Quat &q, const Quat &p) { return dot4_neon((float*)&q.x, (float*)&p.x); }
 inline Quat scale(const Quat &q, float32 r) { return makeQuat(q.w*r, q.x*r, q.y*r, q.z*r); }
 inline float32 length(const Quat &q) { return sqrtf(q.w*q.w + q.x*q.x + q.y*q.y + q.z*q.z); }
-inline Quat normalize(const Quat &q) { return scale(q, 1.0f/length(q)); }
+inline Quat normalize(const Quat &q) { float r[4]; normalize4_neon((float*)&q.x, r); return makeQuat(r[0], r[1], r[2], r[3]); }
 inline Quat conj(const Quat &q) { return makeQuat(q.w, -q.x, -q.y, -q.z); }
 Quat mult(const Quat &q, const Quat &p);
 inline V3d rotate(const V3d &v, const Quat &q) { return mult(mult(q, makeQuat(0.0f, v)), conj(q)).vec(); }