Merge "Add matrix ops to RSH headers."

7 files changed, 499 insertions, 300 deletions

diff --git a/graphics/java/android/renderscript/FieldPacker.java b/graphics/java/android/renderscript/FieldPacker.java
index d166972..f03b51c 100644
--- a/graphics/java/android/renderscript/FieldPacker.java
+++ b/graphics/java/android/renderscript/FieldPacker.java
@@ -244,7 +244,7 @@ public class FieldPacker {
         addU32(v.w);
     }
 
-    public void addBoolean(Boolean v) {
+    public void addBoolean(boolean v) {
         addI8((byte)(v ? 1 : 0));
     }
 
diff --git a/libs/rs/java/ImageProcessing/res/raw/threshold.rs b/libs/rs/java/ImageProcessing/res/raw/threshold.rs
index a8eb164..9585e92 100644
--- a/libs/rs/java/ImageProcessing/res/raw/threshold.rs
+++ b/libs/rs/java/ImageProcessing/res/raw/threshold.rs
@@ -4,6 +4,9 @@
 #include "../../../../scriptc/rs_math.rsh"
 #include "../../../../scriptc/rs_graphics.rsh"
 
+#pragma rs java_package_name(com.android.rs.image)
+
+
 #define MAX_RADIUS 25
 
 int height;
@@ -32,7 +35,7 @@ static float overInWMinInB;
 
 // Store our coefficients here
 static float gaussian[MAX_RADIUS * 2 + 1];
-static float colorMat[4][4];
+static rs_matrix3x3 colorMat;
 
 static void computeColorMatrix() {
     // Saturation
@@ -48,17 +51,15 @@ static void computeColorMatrix() {
 
     float oneMinusS = 1.0f - saturation;
 
-    rsMatrixLoadIdentity((rs_matrix4x4 *)colorMat);
-
-    colorMat[0][0] = oneMinusS * rWeight + saturation;
-    colorMat[0][1] = oneMinusS * rWeight;
-    colorMat[0][2] = oneMinusS * rWeight;
-    colorMat[1][0] = oneMinusS * gWeight;
-    colorMat[1][1] = oneMinusS * gWeight + saturation;
-    colorMat[1][2] = oneMinusS * gWeight;
-    colorMat[2][0] = oneMinusS * bWeight;
-    colorMat[2][1] = oneMinusS * bWeight;
-    colorMat[2][2] = oneMinusS * bWeight + saturation;
+    rsMatrixSet(&colorMat, 0, 0, oneMinusS * rWeight + saturation);
+    rsMatrixSet(&colorMat, 0, 1, oneMinusS * rWeight);
+    rsMatrixSet(&colorMat, 0, 2, oneMinusS * rWeight);
+    rsMatrixSet(&colorMat, 1, 0, oneMinusS * gWeight);
+    rsMatrixSet(&colorMat, 1, 1, oneMinusS * gWeight + saturation);
+    rsMatrixSet(&colorMat, 1, 2, oneMinusS * gWeight);
+    rsMatrixSet(&colorMat, 2, 0, oneMinusS * bWeight);
+    rsMatrixSet(&colorMat, 2, 1, oneMinusS * bWeight);
+    rsMatrixSet(&colorMat, 2, 2, oneMinusS * bWeight + saturation);
 
     inWMinInB = inWhite - inBlack;
     outWMinOutB = outWhite - outBlack;
@@ -107,48 +108,10 @@ static void computeGaussianWeights() {
 
 // This needs to be inline
 static float4 levelsSaturation(float4 currentPixel) {
-#if 0
-    // Color matrix multiply
-    float tempX = colorMat[0][0] * currentPixel.x + colorMat[1][0] * currentPixel.y + colorMat[2][0] * currentPixel.z;
-    float tempY = colorMat[0][1] * currentPixel.x + colorMat[1][1] * currentPixel.y + colorMat[2][1] * currentPixel.z;
-    float tempZ = colorMat[0][2] * currentPixel.x + colorMat[1][2] * currentPixel.y + colorMat[2][2] * currentPixel.z;
-
-    currentPixel.x = tempX;
-    currentPixel.y = tempY;
-    currentPixel.z = tempZ;
-
-    // Clamp to 0..255
-    // Inline the code here to avoid funciton calls
-    currentPixel.x = currentPixel.x > 255.0f ? 255.0f : currentPixel.x;
-    currentPixel.y = currentPixel.y > 255.0f ? 255.0f : currentPixel.y;
-    currentPixel.z = currentPixel.z > 255.0f ? 255.0f : currentPixel.z;
-
-    currentPixel.x = currentPixel.x <= 0.0f ? 0.1f : currentPixel.x;
-    currentPixel.y = currentPixel.y <= 0.0f ? 0.1f : currentPixel.y;
-    currentPixel.z = currentPixel.z <= 0.0f ? 0.1f : currentPixel.z;
-
-    currentPixel.x = pow( (currentPixel.x - inBlack) * overInWMinInB, gamma) * outWMinOutB + outBlack;
-    currentPixel.y = pow( (currentPixel.y - inBlack) * overInWMinInB, gamma) * outWMinOutB + outBlack;
-    currentPixel.z = pow( (currentPixel.z - inBlack) * overInWMinInB, gamma) * outWMinOutB + outBlack;
-
-    currentPixel.x = currentPixel.x > 255.0f ? 255.0f : currentPixel.x;
-    currentPixel.y = currentPixel.y > 255.0f ? 255.0f : currentPixel.y;
-    currentPixel.z = currentPixel.z > 255.0f ? 255.0f : currentPixel.z;
-
-    currentPixel.x = currentPixel.x <= 0.0f ? 0.1f : currentPixel.x;
-    currentPixel.y = currentPixel.y <= 0.0f ? 0.1f : currentPixel.y;
-    currentPixel.z = currentPixel.z <= 0.0f ? 0.1f : currentPixel.z;
-#else
-    float3 temp;
-    // Color matrix multiply
-    temp.x = colorMat[0][0] * currentPixel.x + colorMat[1][0] * currentPixel.y + colorMat[2][0] * currentPixel.z;
-    temp.y = colorMat[0][1] * currentPixel.x + colorMat[1][1] * currentPixel.y + colorMat[2][1] * currentPixel.z;
-    temp.z = colorMat[0][2] * currentPixel.x + colorMat[1][2] * currentPixel.y + colorMat[2][2] * currentPixel.z;
+    float3 temp = rsMatrixMultiply(&colorMat, currentPixel.xyz);
     temp = (clamp(temp, 0.1f, 255.f) - inBlack) * overInWMinInB;
     temp = pow(temp, (float3)gamma);
     currentPixel.xyz = clamp(temp * outWMinOutB + outBlack, 0.1f, 255.f);
-#endif
-
     return currentPixel;
 }
 
@@ -188,8 +151,10 @@ static void horizontalBlur() {
     // Horizontal blur
     int w, h, r;
     for(h = 0; h < height; h ++) {
-        for(w = 0; w < width; w ++) {
+        uchar4 *input = InPixel + h*width;
+        uchar4 *output = ScratchPixel + h*width;
 
+        for(w = 0; w < width; w ++) {
             blurredPixel = 0;
 
             for(r = -radius; r <= radius; r ++) {
@@ -202,23 +167,22 @@ static void horizontalBlur() {
                 if(validW > width - 1) {
                     validW = width - 1;
                 }
-
-                uchar4 *input = InPixel + h*width + validW;
+                //int validW = rsClamp(w + r, 0, width - 1);
 
                 float weight = gaussian[r + radius];
-                currentPixel.x = (float)(input->x);
-                currentPixel.y = (float)(input->y);
-                currentPixel.z = (float)(input->z);
+                currentPixel.x = (float)(input[validW].x);
+                currentPixel.y = (float)(input[validW].y);
+                currentPixel.z = (float)(input[validW].z);
                 //currentPixel.w = (float)(input->a);
 
                 blurredPixel += currentPixel*weight;
             }
 
-            uchar4 *output = ScratchPixel + h*width + w;
             output->x = (uint8_t)blurredPixel.x;
             output->y = (uint8_t)blurredPixel.y;
             output->z = (uint8_t)blurredPixel.z;
             //output->a = (uint8_t)blurredPixel.w;
+            output++;
         }
     }
 }
@@ -229,8 +193,9 @@ static void horizontalBlurLevels() {
     // Horizontal blur
     int w, h, r;
     for(h = 0; h < height; h ++) {
-        for(w = 0; w < width; w ++) {
+        uchar4 *output = OutPixel + h*width;
 
+        for(w = 0; w < width; w ++) {
             blurredPixel = 0;
 
             for(r = -radius; r <= radius; r ++) {
@@ -243,6 +208,7 @@ static void horizontalBlurLevels() {
                 if(validW > width - 1) {
                     validW = width - 1;
                 }
+                //int validW = rsClamp(w + r, 0, width - 1);
 
                 uchar4 *input = InPixel + h*width + validW;
 
@@ -252,16 +218,16 @@ static void horizontalBlurLevels() {
                 currentPixel.z = (float)(input->z);
                 //currentPixel.w = (float)(input->a);
 
-                blurredPixel += currentPixel*weight;
+                blurredPixel.xyz += currentPixel.xyz * weight;
             }
 
             blurredPixel = levelsSaturation(blurredPixel);
 
-            uchar4 *output = ScratchPixel + h*width + w;
             output->x = (uint8_t)blurredPixel.x;
             output->y = (uint8_t)blurredPixel.y;
             output->z = (uint8_t)blurredPixel.z;
             //output->a = (uint8_t)blurredPixel.w;
+            output++;
         }
     }
 }
@@ -272,10 +238,13 @@ static void verticalBlur() {
     // Vertical blur
     int w, h, r;
     for(h = 0; h < height; h ++) {
+        uchar4 *output = OutPixel + h*width;
+
         for(w = 0; w < width; w ++) {
 
             blurredPixel = 0;
             for(r = -radius; r <= radius; r ++) {
+#if 1
                 int validH = h + r;
                 // Clamp to zero and width
                 if(validH < 0) {
@@ -292,17 +261,21 @@ static void verticalBlur() {
                 currentPixel.x = (float)(input->x);
                 currentPixel.y = (float)(input->y);
                 currentPixel.z = (float)(input->z);
-                //currentPixel.w = (float)(input->a);
 
-                blurredPixel += currentPixel*weight;
+                blurredPixel.xyz += currentPixel.xyz * weight;
+#else
+                int validH = rsClamp(h + r, 0, height - 1);
+                uchar4 *input = ScratchPixel + validH*width + w;
+                blurredPixel.xyz += convert_float3(input->xyz) * gaussian[r + radius];
+#endif
             }
 
-            uchar4 *output = OutPixel + h*width + w;
-
+            //output->xyz = convert_uchar3(blurredPixel.xyz);
             output->x = (uint8_t)blurredPixel.x;
             output->y = (uint8_t)blurredPixel.y;
             output->z = (uint8_t)blurredPixel.z;
             //output->a = (uint8_t)blurredPixel.w;
+            output++;
         }
     }
 }
@@ -311,12 +284,6 @@ void filter() {
     RS_DEBUG(height);
     RS_DEBUG(width);
     RS_DEBUG(radius);
-    RS_DEBUG(inBlack);
-    RS_DEBUG(outBlack);
-    RS_DEBUG(inWhite);
-    RS_DEBUG(outWhite);
-    RS_DEBUG(gamma);
-    RS_DEBUG(saturation);
 
     computeColorMatrix();
 
diff --git a/libs/rs/java/ImageProcessing/res/raw/threshold_bc.bc b/libs/rs/java/ImageProcessing/res/raw/threshold_bc.bc
index fd60f76..95dcd8d 100644
--- a/libs/rs/java/ImageProcessing/res/raw/threshold_bc.bc
+++ b/libs/rs/java/ImageProcessing/res/raw/threshold_bc.bc
diff --git a/libs/rs/rsScriptC_Lib.cpp b/libs/rs/rsScriptC_Lib.cpp
index 3f3ff23..ea01134 100644
--- a/libs/rs/rsScriptC_Lib.cpp
+++ b/libs/rs/rsScriptC_Lib.cpp
@@ -32,133 +32,6 @@ using namespace android::renderscript;
     ScriptC * sc = (ScriptC *) tls->mScript
 
 
-
-
-//////////////////////////////////////////////////////////////////////////////
-// Non-Updated code below
-//////////////////////////////////////////////////////////////////////////////
-
-typedef struct {
-    float x;
-    float y;
-    float z;
-} vec3_t;
-
-typedef struct {
-    float x;
-    float y;
-    float z;
-    float w;
-} vec4_t;
-
-typedef struct {
-    float x;
-    float y;
-} vec2_t;
-
-
-//////////////////////////////////////////////////////////////////////////////
-// Vec3 routines
-//////////////////////////////////////////////////////////////////////////////
-
-static void SC_vec3Norm(vec3_t *v)
-{
-    float len = sqrtf(v->x * v->x + v->y * v->y + v->z * v->z);
-    len = 1 / len;
-    v->x *= len;
-    v->y *= len;
-    v->z *= len;
-}
-
-static float SC_vec3Length(const vec3_t *v)
-{
-    return sqrtf(v->x * v->x + v->y * v->y + v->z * v->z);
-}
-
-static void SC_vec3Add(vec3_t *dest, const vec3_t *lhs, const vec3_t *rhs)
-{
-    dest->x = lhs->x + rhs->x;
-    dest->y = lhs->y + rhs->y;
-    dest->z = lhs->z + rhs->z;
-}
-
-static void SC_vec3Sub(vec3_t *dest, const vec3_t *lhs, const vec3_t *rhs)
-{
-    dest->x = lhs->x - rhs->x;
-    dest->y = lhs->y - rhs->y;
-    dest->z = lhs->z - rhs->z;
-}
-
-static void SC_vec3Cross(vec3_t *dest, const vec3_t *lhs, const vec3_t *rhs)
-{
-    float x = lhs->y * rhs->z  - lhs->z * rhs->y;
-    float y = lhs->z * rhs->x  - lhs->x * rhs->z;
-    float z = lhs->x * rhs->y  - lhs->y * rhs->x;
-    dest->x = x;
-    dest->y = y;
-    dest->z = z;
-}
-
-static float SC_vec3Dot(const vec3_t *lhs, const vec3_t *rhs)
-{
-    return lhs->x * rhs->x + lhs->y * rhs->y + lhs->z * rhs->z;
-}
-
-static void SC_vec3Scale(vec3_t *lhs, float scale)
-{
-    lhs->x *= scale;
-    lhs->y *= scale;
-    lhs->z *= scale;
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// Vec4 routines
-//////////////////////////////////////////////////////////////////////////////
-
-static void SC_vec4Norm(vec4_t *v)
-{
-    float len = sqrtf(v->x * v->x + v->y * v->y + v->z * v->z + v->w * v->w);
-    len = 1 / len;
-    v->x *= len;
-    v->y *= len;
-    v->z *= len;
-    v->w *= len;
-}
-
-static float SC_vec4Length(const vec4_t *v)
-{
-    return sqrtf(v->x * v->x + v->y * v->y + v->z * v->z + v->w * v->w);
-}
-
-static void SC_vec4Add(vec4_t *dest, const vec4_t *lhs, const vec4_t *rhs)
-{
-    dest->x = lhs->x + rhs->x;
-    dest->y = lhs->y + rhs->y;
-    dest->z = lhs->z + rhs->z;
-    dest->w = lhs->w + rhs->w;
-}
-
-static void SC_vec4Sub(vec4_t *dest, const vec4_t *lhs, const vec4_t *rhs)
-{
-    dest->x = lhs->x - rhs->x;
-    dest->y = lhs->y - rhs->y;
-    dest->z = lhs->z - rhs->z;
-    dest->w = lhs->w - rhs->w;
-}
-
-static float SC_vec4Dot(const vec4_t *lhs, const vec4_t *rhs)
-{
-    return lhs->x * rhs->x + lhs->y * rhs->y + lhs->z * rhs->z + lhs->w * rhs->w;
-}
-
-static void SC_vec4Scale(vec4_t *lhs, float scale)
-{
-    lhs->x *= scale;
-    lhs->y *= scale;
-    lhs->z *= scale;
-    lhs->w *= scale;
-}
-
 //////////////////////////////////////////////////////////////////////////////
 // Math routines
 //////////////////////////////////////////////////////////////////////////////
@@ -224,41 +97,6 @@ static int SC_randi2(int min, int max)
     return (int)SC_randf2(min, max);
 }
 
-static int SC_clamp(int amount, int low, int high)
-{
-    return amount < low ? low : (amount > high ? high : amount);
-}
-
-static float SC_roundf(float v)
-{
-    return floorf(v + 0.4999999999);
-}
-
-static float SC_distf2(float x1, float y1, float x2, float y2)
-{
-    float x = x2 - x1;
-    float y = y2 - y1;
-    return sqrtf(x * x + y * y);
-}
-
-static float SC_distf3(float x1, float y1, float z1, float x2, float y2, float z2)
-{
-    float x = x2 - x1;
-    float y = y2 - y1;
-    float z = z2 - z1;
-    return sqrtf(x * x + y * y + z * z);
-}
-
-static float SC_magf2(float a, float b)
-{
-    return sqrtf(a * a + b * b);
-}
-
-static float SC_magf3(float a, float b, float c)
-{
-    return sqrtf(a * a + b * b + c * c);
-}
-
 static float SC_frac(float v)
 {
     int i = (int)floor(v);
@@ -512,24 +350,6 @@ static void SC_debugI32(const char *s, int32_t i) {
     LOGE("%s %i  0x%x", s, i, i);
 }
 
-static uchar4 SC_convertColorTo8888_f3(float r, float g, float b) {
-    uchar4 t;
-    t.f[0] = (uint8_t)(r * 255.f);
-    t.f[1] = (uint8_t)(g * 255.f);
-    t.f[2] = (uint8_t)(b * 255.f);
-    t.f[3] = 0xff;
-    return t;
-}
-
-static uchar4 SC_convertColorTo8888_f4(float r, float g, float b, float a) {
-    uchar4 t;
-    t.f[0] = (uint8_t)(r * 255.f);
-    t.f[1] = (uint8_t)(g * 255.f);
-    t.f[2] = (uint8_t)(b * 255.f);
-    t.f[3] = (uint8_t)(a * 255.f);
-    return t;
-}
-
 static uint32_t SC_toClient(void *data, int cmdID, int len, int waitForSpace)
 {
     GET_TLS();
@@ -640,16 +460,6 @@ static ScriptCState::SymbolTable_t gSyms[] = {
     { "rsAllocationGetDimFaces", (void *)&SC_allocGetDimFaces },
     { "rsGetAllocation", (void *)&SC_getAllocation },
 
-    // color
-    { "_Z17rsPackColorTo8888fff", (void *)&SC_convertColorTo8888_f3 },
-    { "_Z17rsPackColorTo8888ffff", (void *)&SC_convertColorTo8888_f4 },
-    //extern uchar4 __attribute__((overloadable)) rsPackColorTo8888(float3);
-    //extern uchar4 __attribute__((overloadable)) rsPackColorTo8888(float4);
-    //extern float4 rsUnpackColor8888(uchar4);
-    //extern uchar4 __attribute__((overloadable)) rsPackColorTo565(float r, float g, float b);
-    //extern uchar4 __attribute__((overloadable)) rsPackColorTo565(float3);
-    //extern float4 rsUnpackColor565(uchar4);
-
     // Debug
     { "_Z7rsDebugPKcf", (void *)&SC_debugF },
     { "_Z7rsDebugPKcDv2_f", (void *)&SC_debugFv2 },
@@ -703,12 +513,6 @@ static ScriptCState::SymbolTable_t gSyms[] = {
 
     //{ "sinf_fast", (void *)&SC_sinf_fast },
     //{ "cosf_fast", (void *)&SC_cosf_fast },
-    //{ "clamp", (void *)&SC_clamp },
-    //{ "distf2", (void *)&SC_distf2 },
-    //{ "distf3", (void *)&SC_distf3 },
-    //{ "magf2", (void *)&SC_magf2 },
-    //{ "magf3", (void *)&SC_magf3 },
-    //{ "mapf", (void *)&SC_mapf },
 
     { "scriptCall", (void *)&SC_scriptCall },
 
diff --git a/libs/rs/scriptc/rs_core.rsh b/libs/rs/scriptc/rs_core.rsh
index ebf6876..91b30b9 100644
--- a/libs/rs/scriptc/rs_core.rsh
+++ b/libs/rs/scriptc/rs_core.rsh
@@ -48,6 +48,457 @@ static float4 rsUnpackColor8888(uchar4 c)
 //extern float4 rsUnpackColor565(uchar4);
 
 
+/////////////////////////////////////////////////////
+// Matrix ops
+/////////////////////////////////////////////////////
+
+static void __attribute__((overloadable))
+rsMatrixSet(rs_matrix4x4 *m, uint32_t row, uint32_t col, float v) {
+    m->m[row * 4 + col] = v;
+}
+
+static float __attribute__((overloadable))
+rsMatrixGet(const rs_matrix4x4 *m, uint32_t row, uint32_t col) {
+    return m->m[row * 4 + col];
+}
+
+static void __attribute__((overloadable))
+rsMatrixSet(rs_matrix3x3 *m, uint32_t row, uint32_t col, float v) {
+    m->m[row * 3 + col] = v;
+}
+
+static float __attribute__((overloadable))
+rsMatrixGet(const rs_matrix3x3 *m, uint32_t row, uint32_t col) {
+    return m->m[row * 3 + col];
+}
+
+static void __attribute__((overloadable))
+rsMatrixSet(rs_matrix2x2 *m, uint32_t row, uint32_t col, float v) {
+    m->m[row * 2 + col] = v;
+}
+
+static float __attribute__((overloadable))
+rsMatrixGet(const rs_matrix2x2 *m, uint32_t row, uint32_t col) {
+    return m->m[row * 2 + col];
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadIdentity(rs_matrix4x4 *m) {
+    m->m[0] = 1.f;
+    m->m[1] = 0.f;
+    m->m[2] = 0.f;
+    m->m[3] = 0.f;
+    m->m[4] = 0.f;
+    m->m[5] = 1.f;
+    m->m[6] = 0.f;
+    m->m[7] = 0.f;
+    m->m[8] = 0.f;
+    m->m[9] = 0.f;
+    m->m[10] = 1.f;
+    m->m[11] = 0.f;
+    m->m[12] = 0.f;
+    m->m[13] = 0.f;
+    m->m[14] = 0.f;
+    m->m[15] = 1.f;
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadIdentity(rs_matrix3x3 *m) {
+    m->m[0] = 1.f;
+    m->m[1] = 0.f;
+    m->m[2] = 0.f;
+    m->m[3] = 0.f;
+    m->m[4] = 1.f;
+    m->m[5] = 0.f;
+    m->m[6] = 0.f;
+    m->m[7] = 0.f;
+    m->m[8] = 1.f;
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadIdentity(rs_matrix2x2 *m) {
+    m->m[0] = 1.f;
+    m->m[1] = 0.f;
+    m->m[2] = 0.f;
+    m->m[3] = 1.f;
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoad(rs_matrix4x4 *m, const float *v) {
+    m->m[0] = v[0];
+    m->m[1] = v[1];
+    m->m[2] = v[2];
+    m->m[3] = v[3];
+    m->m[4] = v[4];
+    m->m[5] = v[5];
+    m->m[6] = v[6];
+    m->m[7] = v[7];
+    m->m[8] = v[8];
+    m->m[9] = v[9];
+    m->m[10] = v[10];
+    m->m[11] = v[11];
+    m->m[12] = v[12];
+    m->m[13] = v[13];
+    m->m[14] = v[14];
+    m->m[15] = v[15];
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoad(rs_matrix3x3 *m, const float *v) {
+    m->m[0] = v[0];
+    m->m[1] = v[1];
+    m->m[2] = v[2];
+    m->m[3] = v[3];
+    m->m[4] = v[4];
+    m->m[5] = v[5];
+    m->m[6] = v[6];
+    m->m[7] = v[7];
+    m->m[8] = v[8];
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoad(rs_matrix2x2 *m, const float *v) {
+    m->m[0] = v[0];
+    m->m[1] = v[1];
+    m->m[2] = v[2];
+    m->m[3] = v[3];
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoad(rs_matrix4x4 *m, const rs_matrix4x4 *v) {
+    m->m[0] = v->m[0];
+    m->m[1] = v->m[1];
+    m->m[2] = v->m[2];
+    m->m[3] = v->m[3];
+    m->m[4] = v->m[4];
+    m->m[5] = v->m[5];
+    m->m[6] = v->m[6];
+    m->m[7] = v->m[7];
+    m->m[8] = v->m[8];
+    m->m[9] = v->m[9];
+    m->m[10] = v->m[10];
+    m->m[11] = v->m[11];
+    m->m[12] = v->m[12];
+    m->m[13] = v->m[13];
+    m->m[14] = v->m[14];
+    m->m[15] = v->m[15];
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoad(rs_matrix4x4 *m, const rs_matrix3x3 *v) {
+    m->m[0] = v->m[0];
+    m->m[1] = v->m[1];
+    m->m[2] = v->m[2];
+    m->m[3] = 0.f;
+    m->m[4] = v->m[3];
+    m->m[5] = v->m[4];
+    m->m[6] = v->m[5];
+    m->m[7] = 0.f;
+    m->m[8] = v->m[6];
+    m->m[9] = v->m[7];
+    m->m[10] = v->m[8];
+    m->m[11] = 0.f;
+    m->m[12] = 0.f;
+    m->m[13] = 0.f;
+    m->m[14] = 0.f;
+    m->m[15] = 1.f;
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoad(rs_matrix4x4 *m, const rs_matrix2x2 *v) {
+    m->m[0] = v->m[0];
+    m->m[1] = v->m[1];
+    m->m[2] = 0.f;
+    m->m[3] = 0.f;
+    m->m[4] = v->m[3];
+    m->m[5] = v->m[4];
+    m->m[6] = 0.f;
+    m->m[7] = 0.f;
+    m->m[8] = v->m[6];
+    m->m[9] = v->m[7];
+    m->m[10] = 1.f;
+    m->m[11] = 0.f;
+    m->m[12] = 0.f;
+    m->m[13] = 0.f;
+    m->m[14] = 0.f;
+    m->m[15] = 1.f;
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoad(rs_matrix3x3 *m, const rs_matrix3x3 *v) {
+    m->m[0] = v->m[0];
+    m->m[1] = v->m[1];
+    m->m[2] = v->m[2];
+    m->m[3] = v->m[3];
+    m->m[4] = v->m[4];
+    m->m[5] = v->m[5];
+    m->m[6] = v->m[6];
+    m->m[7] = v->m[7];
+    m->m[8] = v->m[8];
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoad(rs_matrix2x2 *m, const rs_matrix2x2 *v) {
+    m->m[0] = v->m[0];
+    m->m[1] = v->m[1];
+    m->m[2] = v->m[2];
+    m->m[3] = v->m[3];
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadRotate(rs_matrix4x4 *m, float rot, float x, float y, float z) {
+    float c, s;
+    m->m[3] = 0;
+    m->m[7] = 0;
+    m->m[11]= 0;
+    m->m[12]= 0;
+    m->m[13]= 0;
+    m->m[14]= 0;
+    m->m[15]= 1;
+    rot *= (float)(M_PI / 180.0f);
+    c = cos(rot);
+    s = sin(rot);
+
+    const float len = x*x + y*y + z*z;
+    if (!(len != 1)) {
+        const float recipLen = 1.f / sqrt(len);
+        x *= recipLen;
+        y *= recipLen;
+        z *= recipLen;
+    }
+    const float nc = 1.0f - c;
+    const float xy = x * y;
+    const float yz = y * z;
+    const float zx = z * x;
+    const float xs = x * s;
+    const float ys = y * s;
+    const float zs = z * s;
+    m->m[ 0] = x*x*nc +  c;
+    m->m[ 4] =  xy*nc - zs;
+    m->m[ 8] =  zx*nc + ys;
+    m->m[ 1] =  xy*nc + zs;
+    m->m[ 5] = y*y*nc +  c;
+    m->m[ 9] =  yz*nc - xs;
+    m->m[ 2] =  zx*nc - ys;
+    m->m[ 6] =  yz*nc + xs;
+    m->m[10] = z*z*nc +  c;
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadScale(rs_matrix4x4 *m, float x, float y, float z) {
+    rsMatrixLoadIdentity(m);
+    m->m[0] = x;
+    m->m[5] = y;
+    m->m[10] = z;
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadTranslate(rs_matrix4x4 *m, float x, float y, float z) {
+    rsMatrixLoadIdentity(m);
+    m->m[12] = x;
+    m->m[13] = y;
+    m->m[14] = z;
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadMultiply(rs_matrix4x4 *m, const rs_matrix4x4 *lhs, const rs_matrix4x4 *rhs) {
+    for (int i=0 ; i<4 ; i++) {
+        float ri0 = 0;
+        float ri1 = 0;
+        float ri2 = 0;
+        float ri3 = 0;
+        for (int j=0 ; j<4 ; j++) {
+            const float rhs_ij = rsMatrixGet(rhs, i,j);
+            ri0 += rsMatrixGet(lhs, j, 0) * rhs_ij;
+            ri1 += rsMatrixGet(lhs, j, 1) * rhs_ij;
+            ri2 += rsMatrixGet(lhs, j, 2) * rhs_ij;
+            ri3 += rsMatrixGet(lhs, j, 3) * rhs_ij;
+        }
+        rsMatrixSet(m, i, 0, ri0);
+        rsMatrixSet(m, i, 1, ri1);
+        rsMatrixSet(m, i, 2, ri2);
+        rsMatrixSet(m, i, 3, ri3);
+    }
+}
+
+static void __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix4x4 *m, const rs_matrix4x4 *rhs) {
+    rs_matrix4x4 mt;
+    rsMatrixLoadMultiply(&mt, m, rhs);
+    rsMatrixLoad(m, &mt);
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadMultiply(rs_matrix3x3 *m, const rs_matrix3x3 *lhs, const rs_matrix3x3 *rhs) {
+    for (int i=0 ; i<3 ; i++) {
+        float ri0 = 0;
+        float ri1 = 0;
+        float ri2 = 0;
+        for (int j=0 ; j<3 ; j++) {
+            const float rhs_ij = rsMatrixGet(rhs, i,j);
+            ri0 += rsMatrixGet(lhs, j, 0) * rhs_ij;
+            ri1 += rsMatrixGet(lhs, j, 1) * rhs_ij;
+            ri2 += rsMatrixGet(lhs, j, 2) * rhs_ij;
+        }
+        rsMatrixSet(m, i, 0, ri0);
+        rsMatrixSet(m, i, 1, ri1);
+        rsMatrixSet(m, i, 2, ri2);
+    }
+}
+
+static void __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix3x3 *m, const rs_matrix3x3 *rhs) {
+    rs_matrix3x3 mt;
+    rsMatrixLoadMultiply(&mt, m, rhs);
+    rsMatrixLoad(m, &mt);
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadMultiply(rs_matrix2x2 *m, const rs_matrix2x2 *lhs, const rs_matrix2x2 *rhs) {
+    for (int i=0 ; i<2 ; i++) {
+        float ri0 = 0;
+        float ri1 = 0;
+        for (int j=0 ; j<2 ; j++) {
+            const float rhs_ij = rsMatrixGet(rhs, i,j);
+            ri0 += rsMatrixGet(lhs, j, 0) * rhs_ij;
+            ri1 += rsMatrixGet(lhs, j, 1) * rhs_ij;
+        }
+        rsMatrixSet(m, i, 0, ri0);
+        rsMatrixSet(m, i, 1, ri1);
+    }
+}
+
+static void __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix2x2 *m, const rs_matrix2x2 *rhs) {
+    rs_matrix2x2 mt;
+    rsMatrixLoadMultiply(&mt, m, rhs);
+    rsMatrixLoad(m, &mt);
+}
+
+static void __attribute__((overloadable))
+rsMatrixRotate(rs_matrix4x4 *m, float rot, float x, float y, float z) {
+    rs_matrix4x4 m1;
+    rsMatrixLoadRotate(&m1, rot, x, y, z);
+    rsMatrixMultiply(m, &m1);
+}
+
+static void __attribute__((overloadable))
+rsMatrixScale(rs_matrix4x4 *m, float x, float y, float z) {
+    rs_matrix4x4 m1;
+    rsMatrixLoadScale(&m1, x, y, z);
+    rsMatrixMultiply(m, &m1);
+}
+
+static void __attribute__((overloadable))
+rsMatrixTranslate(rs_matrix4x4 *m, float x, float y, float z) {
+    rs_matrix4x4 m1;
+    rsMatrixLoadTranslate(&m1, x, y, z);
+    rsMatrixMultiply(m, &m1);
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadOrtho(rs_matrix4x4 *m, float left, float right, float bottom, float top, float near, float far) {
+    rsMatrixLoadIdentity(m);
+    m->m[0] = 2.f / (right - left);
+    m->m[5] = 2.f / (top - bottom);
+    m->m[10]= -2.f / (far - near);
+    m->m[12]= -(right + left) / (right - left);
+    m->m[13]= -(top + bottom) / (top - bottom);
+    m->m[14]= -(far + near) / (far - near);
+}
+
+static void __attribute__((overloadable))
+rsMatrixLoadFrustum(rs_matrix4x4 *m, float left, float right, float bottom, float top, float near, float far) {
+    rsMatrixLoadIdentity(m);
+    m->m[0] = 2.f * near / (right - left);
+    m->m[5] = 2.f * near / (top - bottom);
+    m->m[8] = (right + left) / (right - left);
+    m->m[9] = (top + bottom) / (top - bottom);
+    m->m[10]= -(far + near) / (far - near);
+    m->m[11]= -1.f;
+    m->m[14]= -2.f * far * near / (far - near);
+    m->m[15]= 0.f;
+}
+
+static float4 __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix4x4 *m, float4 in) {
+    float4 ret;
+    ret.x = (m->m[0] * in.x) + (m->m[4] * in.y) + (m->m[8] * in.z) + (m->m[12] * in.w);
+    ret.y = (m->m[1] * in.x) + (m->m[5] * in.y) + (m->m[9] * in.z) + (m->m[13] * in.w);
+    ret.z = (m->m[2] * in.x) + (m->m[6] * in.y) + (m->m[10] * in.z) + (m->m[14] * in.w);
+    ret.w = (m->m[3] * in.x) + (m->m[7] * in.y) + (m->m[11] * in.z) + (m->m[15] * in.w);
+    return ret;
+}
+
+static float4 __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix4x4 *m, float3 in) {
+    float4 ret;
+    ret.x = (m->m[0] * in.x) + (m->m[4] * in.y) + (m->m[8] * in.z) + m->m[12];
+    ret.y = (m->m[1] * in.x) + (m->m[5] * in.y) + (m->m[9] * in.z) + m->m[13];
+    ret.z = (m->m[2] * in.x) + (m->m[6] * in.y) + (m->m[10] * in.z) + m->m[14];
+    ret.w = (m->m[3] * in.x) + (m->m[7] * in.y) + (m->m[11] * in.z) + m->m[15];
+    return ret;
+}
+
+static float4 __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix4x4 *m, float2 in) {
+    float4 ret;
+    ret.x = (m->m[0] * in.x) + (m->m[4] * in.y) + m->m[12];
+    ret.y = (m->m[1] * in.x) + (m->m[5] * in.y) + m->m[13];
+    ret.z = (m->m[2] * in.x) + (m->m[6] * in.y) + m->m[14];
+    ret.w = (m->m[3] * in.x) + (m->m[7] * in.y) + m->m[15];
+    return ret;
+}
+
+static float3 __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix3x3 *m, float3 in) {
+    float3 ret;
+    ret.x = (m->m[0] * in.x) + (m->m[3] * in.y) + (m->m[6] * in.z);
+    ret.y = (m->m[1] * in.x) + (m->m[4] * in.y) + (m->m[7] * in.z);
+    ret.z = (m->m[2] * in.x) + (m->m[5] * in.y) + (m->m[8] * in.z);
+    return ret;
+}
+
+static float3 __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix3x3 *m, float2 in) {
+    float3 ret;
+    ret.x = (m->m[0] * in.x) + (m->m[3] * in.y);
+    ret.y = (m->m[1] * in.x) + (m->m[4] * in.y);
+    ret.z = (m->m[2] * in.x) + (m->m[5] * in.y);
+    return ret;
+}
+
+static float2 __attribute__((overloadable))
+rsMatrixMultiply(rs_matrix2x2 *m, float2 in) {
+    float2 ret;
+    ret.x = (m->m[0] * in.x) + (m->m[2] * in.y);
+    ret.y = (m->m[1] * in.x) + (m->m[3] * in.y);
+    return ret;
+}
+
+/////////////////////////////////////////////////////
+// int ops
+/////////////////////////////////////////////////////
+
+__inline__ static float __attribute__((overloadable, always_inline)) rsClamp(uint amount, uint low, uint high) {
+    return amount < low ? low : (amount > high ? high : amount);
+}
+__inline__ static float __attribute__((overloadable, always_inline)) rsClamp(int amount, int low, int high) {
+    return amount < low ? low : (amount > high ? high : amount);
+}
+__inline__ static float __attribute__((overloadable, always_inline)) rsClamp(ushort amount, ushort low, ushort high) {
+    return amount < low ? low : (amount > high ? high : amount);
+}
+__inline__ static float __attribute__((overloadable, always_inline)) rsClamp(short amount, short low, short high) {
+    return amount < low ? low : (amount > high ? high : amount);
+}
+__inline__ static float __attribute__((overloadable, always_inline)) rsClamp(uchar amount, uchar low, uchar high) {
+    return amount < low ? low : (amount > high ? high : amount);
+}
+__inline__ static float __attribute__((overloadable, always_inline)) rsClamp(char amount, char low, char high) {
+    return amount < low ? low : (amount > high ? high : amount);
+}
+
 
 
 #endif
diff --git a/libs/rs/scriptc/rs_math.rsh b/libs/rs/scriptc/rs_math.rsh
index a26491f..8bf53bf 100644
--- a/libs/rs/scriptc/rs_math.rsh
+++ b/libs/rs/scriptc/rs_math.rsh
@@ -44,18 +44,6 @@ extern int64_t rsElapsedTimeMillis();
 
 extern int rsSendToClient(void *data, int cmdID, int len, int waitForSpace);
 
-extern void rsMatrixLoadIdentity(rs_matrix4x4 *mat);
-extern void rsMatrixLoadFloat(rs_matrix4x4 *mat, const float *f);
-extern void rsMatrixLoadMat(rs_matrix4x4 *mat, const rs_matrix4x4 *newmat);
-extern void rsMatrixLoadRotate(rs_matrix4x4 *mat, float rot, float x, float y, float z);
-extern void rsMatrixLoadScale(rs_matrix4x4*mat, float x, float y, float z);
-extern void rsMatrixLoadTranslate(rs_matrix4x4 *mat, float x, float y, float z);
-extern void rsMatrixLoadMultiply(rs_matrix4x4 *mat, const rs_matrix4x4 *lhs, const rs_matrix4x4 *rhs);
-extern void rsMatrixMultiply(rs_matrix4x4 *mat, const rs_matrix4x4 *rhs);
-extern void rsMatrixRotate(rs_matrix4x4 *mat, float rot, float x, float y, float z);
-extern void rsMatrixScale(rs_matrix4x4 *mat, float x, float y, float z);
-extern void rsMatrixTranslate(rs_matrix4x4 *mat, float x, float y, float z);
-
 // Script to Script
 extern void __attribute__((overloadable))rsForEach(rs_script, rs_allocation input);
 extern void __attribute__((overloadable))rsForEach(rs_script, rs_allocation input, rs_allocation output);
@@ -65,23 +53,3 @@ extern void __attribute__((overloadable))rsForEach(rs_script, rs_allocation inpu
 extern void __attribute__((overloadable))rsForEach(rs_script, rs_allocation input, rs_allocation output, int xStart, int yStart, int xEnd, int yEnd);
 
 
-///////////////////////////////////////////////////////////////////
-// non update funcs
-
-/*
-extern float3 float3Norm(float3);
-extern float float3Length(float3);
-extern float3 float3Add(float3 lhs, float3 rhs);
-extern float3 float3Sub(float3 lhs, float3 rhs);
-extern float3 float3Cross(float3 lhs, float3 rhs);
-extern float float3Dot(float3 lhs, float3 rhs);
-extern float3 float3Scale(float3 v, float scale);
-
-extern float4 float4Add(float4 lhs, float4 rhs);
-extern float4 float4Sub(float4 lhs, float4 rhs);
-extern float4 float4Cross(float4 lhs, float4 rhs);
-extern float float4Dot(float4 lhs, float4 rhs);
-extern float4 float4Scale(float4 v, float scale);
-*/
-
-
diff --git a/libs/rs/scriptc/rs_types.rsh b/libs/rs/scriptc/rs_types.rsh
index 465a4a5..1d8f1bd 100644
--- a/libs/rs/scriptc/rs_types.rsh
+++ b/libs/rs/scriptc/rs_types.rsh
@@ -58,5 +58,14 @@ typedef struct {
     float m[16];
 } rs_matrix4x4;
 
+typedef struct {
+    float m[9];
+} rs_matrix3x3;
+
+typedef struct {
+    float m[4];
+} rs_matrix2x2;
+
+
 #define RS_PACKED __attribute__((packed, aligned(4)))