Merge "Clean up RS math fp library functions." into honeycomb

libs/rs/java/tests/src/com/android/rs/test/fp_mad.rs

@@ -102,8 +102,7 @@ static void test_clamp(uint32_t index) {
     start();
 
     // Do ~100 M ops
-    int ct;
+    for (int ct=0; ct < 1000 * 100; ct++) {
-    for (ct=0; ct < 1000 * 100; ct++) {
         for (int i=0; i < (1000); i++) {
             data_f1[i] = clamp(data_f1[i], -1.f, 1.f);
         }
@@ -114,7 +113,7 @@ static void test_clamp(uint32_t index) {
 
     start();
     // Do ~100 M ops
-    for (ct=0; ct < 1000 * 100; ct++) {
+    for (int ct=0; ct < 1000 * 100; ct++) {
         for (int i=0; i < (1000); i++) {
             if (data_f1[i] < -1.f) data_f1[i] = -1.f;
             if (data_f1[i] > -1.f) data_f1[i] = 1.f;
@@ -130,8 +129,7 @@ static void test_clamp4(uint32_t index) {
 
     float total = 0;
     // Do ~100 M ops
-    int ct;
+    for (int ct=0; ct < 1000 * 100 /4; ct++) {
-    for (ct=0; ct < 1000 * 100 /4; ct++) {
         for (int i=0; i < (1000); i++) {
             data_f4[i] = clamp(data_f4[i], -1.f, 1.f);
         }

libs/rs/java/tests/src/com/android/rs/test/math.rs

@@ -7,37 +7,154 @@ volatile float2 f2;
 volatile float3 f3;
 volatile float4 f4;
 
-#define TEST_F(fnc, var)            \
+volatile int i1;
-    rsDebug("Testing " #fnc, 0);    \
+volatile int2 i2;
-    var##1 = fnc(var##1);           \
+volatile int3 i3;
-    var##2 = fnc(var##2);           \
+volatile int4 i4;
-    var##3 = fnc(var##3);           \
-    var##4 = fnc(var##4);
 
-#define TEST_FP(fnc, var)           \
+#define TEST_FN_FUNC_FN(fnc)        \
     rsDebug("Testing " #fnc, 0);    \
-    var##1 = fnc(var##1, (float*) &f1);  \
+    f1 = fnc(f1);                   \
-    var##2 = fnc(var##2, (float2*) &f2);  \
+    f2 = fnc(f2);                   \
-    var##3 = fnc(var##3, (float3*) &f3);  \
+    f3 = fnc(f3);                   \
-    var##4 = fnc(var##4, (float4*) &f4);
+    f4 = fnc(f4);
 
-#define TEST_F2(fnc, var)           \
+#define TEST_FN_FUNC_FN_PFN(fnc)    \
     rsDebug("Testing " #fnc, 0);    \
-    var##1 = fnc(var##1, var##1);   \
+    f1 = fnc(f1, (float*) &f1);     \
-    var##2 = fnc(var##2, var##2);   \
+    f2 = fnc(f2, (float2*) &f2);    \
-    var##3 = fnc(var##3, var##3);   \
+    f3 = fnc(f3, (float3*) &f3);    \
-    var##4 = fnc(var##4, var##4);
+    f4 = fnc(f4, (float4*) &f4);
+
+#define TEST_FN_FUNC_FN_FN(fnc)     \
+    rsDebug("Testing " #fnc, 0);    \
+    f1 = fnc(f1, f1);               \
+    f2 = fnc(f2, f2);               \
+    f3 = fnc(f3, f3);               \
+    f4 = fnc(f4, f4);
+
+#define TEST_FN_FUNC_FN_F(fnc)      \
+    rsDebug("Testing " #fnc, 0);    \
+    f1 = fnc(f1, f1);               \
+    f2 = fnc(f2, f1);               \
+    f3 = fnc(f3, f1);               \
+    f4 = fnc(f4, f1);
+
+#define TEST_FN_FUNC_FN_IN(fnc)     \
+    rsDebug("Testing " #fnc, 0);    \
+    f1 = fnc(f1, i1);               \
+    f2 = fnc(f2, i2);               \
+    f3 = fnc(f3, i3);               \
+    f4 = fnc(f4, i4);
+
+#define TEST_FN_FUNC_FN_I(fnc)      \
+    rsDebug("Testing " #fnc, 0);    \
+    f1 = fnc(f1, i1);               \
+    f2 = fnc(f2, i1);               \
+    f3 = fnc(f3, i1);               \
+    f4 = fnc(f4, i1);
+
+#define TEST_FN_FUNC_FN_FN_FN(fnc)  \
+    rsDebug("Testing " #fnc, 0);    \
+    f1 = fnc(f1, f1, f1);           \
+    f2 = fnc(f2, f2, f2);           \
+    f3 = fnc(f3, f3, f3);           \
+    f4 = fnc(f4, f4, f4);
+
+#define TEST_FN_FUNC_FN_PIN(fnc)    \
+    rsDebug("Testing " #fnc, 0);    \
+    f1 = fnc(f1, (int*) &i1);       \
+    f2 = fnc(f2, (int2*) &i2);      \
+    f3 = fnc(f3, (int3*) &i3);      \
+    f4 = fnc(f4, (int4*) &i4);
+
+#define TEST_FN_FUNC_FN_FN_PIN(fnc) \
+    rsDebug("Testing " #fnc, 0);    \
+    f1 = fnc(f1, f1, (int*) &i1);   \
+    f2 = fnc(f2, f2, (int2*) &i2);  \
+    f3 = fnc(f3, f3, (int3*) &i3);  \
+    f4 = fnc(f4, f4, (int4*) &i4);
+
+#define TEST_IN_FUNC_FN(fnc)        \
+    rsDebug("Testing " #fnc, 0);    \
+    i1 = fnc(f1);                   \
+    i2 = fnc(f2);                   \
+    i3 = fnc(f3);                   \
+    i4 = fnc(f4);
+
 
 static bool test_math(uint32_t index) {
     bool failed = false;
     start();
 
-    TEST_F(cos, f);
+    TEST_FN_FUNC_FN(acos);
-    TEST_FP(modf, f);
+    TEST_FN_FUNC_FN(acosh);
-    TEST_F2(pow, f);
+    TEST_FN_FUNC_FN(acospi);
-    TEST_F(sin, f);
+    TEST_FN_FUNC_FN(asin);
-    TEST_F(sqrt, f);
+    TEST_FN_FUNC_FN(asinh);
-
+    TEST_FN_FUNC_FN(asinpi);
+    TEST_FN_FUNC_FN(atan);
+    TEST_FN_FUNC_FN_FN(atan2);
+    TEST_FN_FUNC_FN(atanh);
+    TEST_FN_FUNC_FN(atanpi);
+    TEST_FN_FUNC_FN_FN(atan2pi);
+    TEST_FN_FUNC_FN(cbrt);
+    TEST_FN_FUNC_FN(ceil);
+    TEST_FN_FUNC_FN_FN(copysign);
+    TEST_FN_FUNC_FN(cos);
+    TEST_FN_FUNC_FN(cosh);
+    TEST_FN_FUNC_FN(cospi);
+    TEST_FN_FUNC_FN(erfc);
+    TEST_FN_FUNC_FN(erf);
+    TEST_FN_FUNC_FN(exp);
+    TEST_FN_FUNC_FN(exp2);
+    TEST_FN_FUNC_FN(exp10);
+    TEST_FN_FUNC_FN(expm1);
+    TEST_FN_FUNC_FN(fabs);
+    TEST_FN_FUNC_FN_FN(fdim);
+    TEST_FN_FUNC_FN(floor);
+    TEST_FN_FUNC_FN_FN_FN(fma);
+    TEST_FN_FUNC_FN_FN(fmax);
+    TEST_FN_FUNC_FN_F(fmax);
+    TEST_FN_FUNC_FN_FN(fmin);
+    TEST_FN_FUNC_FN_F(fmin);
+    TEST_FN_FUNC_FN_FN(fmod);
+    TEST_FN_FUNC_FN_PFN(fract);
+    TEST_FN_FUNC_FN_PIN(frexp);
+    TEST_FN_FUNC_FN_FN(hypot);
+    TEST_IN_FUNC_FN(ilogb);
+    TEST_FN_FUNC_FN_IN(ldexp);
+    TEST_FN_FUNC_FN_I(ldexp);
+    TEST_FN_FUNC_FN(lgamma);
+    TEST_FN_FUNC_FN_PIN(lgamma);
+    TEST_FN_FUNC_FN(log);
+    TEST_FN_FUNC_FN(log2);
+    TEST_FN_FUNC_FN(log10);
+    TEST_FN_FUNC_FN(log1p);
+    TEST_FN_FUNC_FN(logb);
+    TEST_FN_FUNC_FN_FN_FN(mad);
+    TEST_FN_FUNC_FN_PFN(modf);
+    // nan
+    TEST_FN_FUNC_FN_FN(nextafter);
+    TEST_FN_FUNC_FN_FN(pow);
+    TEST_FN_FUNC_FN_IN(pown);
+    TEST_FN_FUNC_FN_FN(powr);
+    TEST_FN_FUNC_FN_FN(remainder);
+    TEST_FN_FUNC_FN_FN_PIN(remquo);
+    TEST_FN_FUNC_FN(rint);
+    TEST_FN_FUNC_FN_IN(rootn);
+    TEST_FN_FUNC_FN(round);
+    TEST_FN_FUNC_FN(rsqrt);
+    TEST_FN_FUNC_FN(sin);
+    TEST_FN_FUNC_FN_PFN(sincos);
+    TEST_FN_FUNC_FN(sinh);
+    TEST_FN_FUNC_FN(sinpi);
+    TEST_FN_FUNC_FN(sqrt);
+    TEST_FN_FUNC_FN(tan);
+    TEST_FN_FUNC_FN(tanh);
+    TEST_FN_FUNC_FN(tanpi);
+    TEST_FN_FUNC_FN(tgamma);
+    TEST_FN_FUNC_FN(trunc);
 
     float time = end(index);
 

libs/rs/rsScriptC_LibCL.cpp

@@ -58,6 +58,10 @@ static float SC_log2(float v) {
     return log10(v) / log10(2.f);
 }
 
+static float SC_mad(float v1, float v2, float v3) {
+    return v1 * v2 + v3;
+}
+
 static float SC_pown(float v, int p) {
     return powf(v, (float)p);
 }
@@ -188,6 +192,7 @@ static ScriptCState::SymbolTable_t gSyms[] = {
     { "_Z6asinpif", (void *)&SC_asinpi, true },
     { "_Z4atanf", (void *)&atanf, true },
     { "_Z5atan2ff", (void *)&atan2f, true },
+    { "_Z5atanhf", (void *)&atanhf, true },
     { "_Z6atanpif", (void *)&SC_atanpi, true },
     { "_Z7atan2piff", (void *)&SC_atan2pi, true },
     { "_Z4cbrtf", (void *)&cbrtf, true },
@@ -215,33 +220,32 @@ static ScriptCState::SymbolTable_t gSyms[] = {
     { "_Z5ilogbf", (void *)&ilogbf, true },
     { "_Z5ldexpfi", (void *)&ldexpf, true },
     { "_Z6lgammaf", (void *)&lgammaf, true },
+    { "_Z6lgammafPi", (void *)&lgammaf_r, true },
     { "_Z3logf", (void *)&logf, true },
     { "_Z4log2f", (void *)&SC_log2, true },
     { "_Z5log10f", (void *)&log10f, true },
     { "_Z5log1pf", (void *)&log1pf, true },
-    //{ "logb", (void *)&, true },
+    { "_Z4logbf", (void *)&logbf, true },
-    //{ "mad", (void *)&, true },
+    { "_Z3madfff", (void *)&SC_mad, true },
     { "_Z4modffPf", (void *)&modff, true },
     //{ "nan", (void *)&, true },
     { "_Z9nextafterff", (void *)&nextafterf, true },
     { "_Z3powff", (void *)&powf, true },
-    { "_Z4pownfi", (void *)&SC_pown, true },
-    { "_Z4powrff", (void *)&SC_powr, true },
     { "_Z9remainderff", (void *)&remainderf, true },
-    { "remquo", (void *)&remquof, true },
+    { "_Z6remquoffPi", (void *)&remquof, true },
     { "_Z4rintf", (void *)&rintf, true },
     { "_Z5rootnfi", (void *)&SC_rootn, true },
     { "_Z5roundf", (void *)&roundf, true },
     { "_Z5rsqrtf", (void *)&SC_rsqrt, true },
     { "_Z3sinf", (void *)&sinf, true },
-    { "sincos", (void *)&SC_sincos, true },
+    { "_Z6sincosfPf", (void *)&SC_sincos, true },
     { "_Z4sinhf", (void *)&sinhf, true },
     { "_Z5sinpif", (void *)&SC_sinpi, true },
     { "_Z4sqrtf", (void *)&sqrtf, true },
     { "_Z3tanf", (void *)&tanf, true },
     { "_Z4tanhf", (void *)&tanhf, true },
     { "_Z5tanpif", (void *)&SC_tanpi, true },
-    //{ "tgamma", (void *)&, true },
+    { "_Z6tgammaf", (void *)&lgammaf, true }, // FIXME!!! NEEDS TO USE tgammaf
     { "_Z5truncf", (void *)&truncf, true },
 
     // OpenCL Int

libs/rs/scriptc/rs_cl.rsh

@@ -42,7 +42,7 @@ CVT_FUNC(float)
 
 // Float ops, 6.11.2
 
-#define DEF_FUNC_1(fnc) \
+#define FN_FUNC_FN(fnc) \
 _RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v) { \
     float2 r; \
     r.x = fnc(v.x); \
@@ -65,7 +65,7 @@ _RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v) { \
     return r; \
 }
 
-#define DEF_FUNC_1_RI(fnc) \
+#define IN_FUNC_FN(fnc) \
 _RS_STATIC int2 __attribute__((overloadable)) fnc(float2 v) { \
     int2 r; \
     r.x = fnc(v.x); \
@@ -88,7 +88,7 @@ _RS_STATIC int4 __attribute__((overloadable)) fnc(float4 v) { \
     return r; \
 }
 
-#define DEF_FUNC_2(fnc) \
+#define FN_FUNC_FN_FN(fnc) \
 _RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v1, float2 v2) { \
     float2 r; \
     r.x = fnc(v1.x, v2.x); \
@@ -111,7 +111,7 @@ _RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v1, float4 v2) { \
     return r; \
 }
 
-#define DEF_FUNC_2F(fnc) \
+#define FN_FUNC_FN_F(fnc) \
 _RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v1, float v2) { \
     float2 r; \
     r.x = fnc(v1.x, v2); \
@@ -134,7 +134,53 @@ _RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v1, float v2) { \
     return r; \
 }
 
-#define DEF_FUNC_2P(fnc) \
+#define FN_FUNC_FN_IN(fnc) \
+_RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v1, int2 v2) { \
+    float2 r; \
+    r.x = fnc(v1.x, v2.x); \
+    r.y = fnc(v1.y, v2.y); \
+    return r; \
+} \
+_RS_STATIC float3 __attribute__((overloadable)) fnc(float3 v1, int3 v2) { \
+    float3 r; \
+    r.x = fnc(v1.x, v2.x); \
+    r.y = fnc(v1.y, v2.y); \
+    r.z = fnc(v1.z, v2.z); \
+    return r; \
+} \
+_RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v1, int4 v2) { \
+    float4 r; \
+    r.x = fnc(v1.x, v2.x); \
+    r.y = fnc(v1.y, v2.y); \
+    r.z = fnc(v1.z, v2.z); \
+    r.w = fnc(v1.w, v2.w); \
+    return r; \
+}
+
+#define FN_FUNC_FN_I(fnc) \
+_RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v1, int v2) { \
+    float2 r; \
+    r.x = fnc(v1.x, v2); \
+    r.y = fnc(v1.y, v2); \
+    return r; \
+} \
+_RS_STATIC float3 __attribute__((overloadable)) fnc(float3 v1, int v2) { \
+    float3 r; \
+    r.x = fnc(v1.x, v2); \
+    r.y = fnc(v1.y, v2); \
+    r.z = fnc(v1.z, v2); \
+    return r; \
+} \
+_RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v1, int v2) { \
+    float4 r; \
+    r.x = fnc(v1.x, v2); \
+    r.y = fnc(v1.y, v2); \
+    r.z = fnc(v1.z, v2); \
+    r.w = fnc(v1.w, v2); \
+    return r; \
+}
+
+#define FN_FUNC_FN_PFN(fnc) \
 _RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v1, float2 *v2) { \
     float2 r; \
     float q; \
@@ -169,112 +215,205 @@ _RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v1, float4 *v2) { \
     return r; \
 }
 
+#define FN_FUNC_FN_PIN(fnc) \
+_RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v1, int2 *v2) { \
+    float2 r; \
+    int q; \
+    r.x = fnc(v1.x, &q); \
+    v2->x = q; \
+    r.y = fnc(v1.y, &q); \
+    v2->y = q; \
+    return r; \
+} \
+_RS_STATIC float3 __attribute__((overloadable)) fnc(float3 v1, int3 *v2) { \
+    float3 r; \
+    int q; \
+    r.x = fnc(v1.x, &q); \
+    v2->x = q; \
+    r.y = fnc(v1.y, &q); \
+    v2->y = q; \
+    r.z = fnc(v1.z, &q); \
+    v2->z = q; \
+    return r; \
+} \
+_RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v1, int4 *v2) { \
+    float4 r; \
+    int q; \
+    r.x = fnc(v1.x, &q); \
+    v2->x = q; \
+    r.y = fnc(v1.y, &q); \
+    v2->y = q; \
+    r.z = fnc(v1.z, &q); \
+    v2->z = q; \
+    r.w = fnc(v1.w, &q); \
+    v2->w = q; \
+    return r; \
+}
+
+#define FN_FUNC_FN_FN_FN(fnc) \
+_RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v1, float2 v2, float2 v3) { \
+    float2 r; \
+    r.x = fnc(v1.x, v2.x, v3.x); \
+    r.y = fnc(v1.y, v2.y, v3.y); \
+    return r; \
+} \
+_RS_STATIC float3 __attribute__((overloadable)) fnc(float3 v1, float3 v2, float3 v3) { \
+    float3 r; \
+    r.x = fnc(v1.x, v2.x, v3.x); \
+    r.y = fnc(v1.y, v2.y, v3.y); \
+    r.z = fnc(v1.z, v2.z, v3.z); \
+    return r; \
+} \
+_RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v1, float4 v2, float4 v3) { \
+    float4 r; \
+    r.x = fnc(v1.x, v2.x, v3.x); \
+    r.y = fnc(v1.y, v2.y, v3.y); \
+    r.z = fnc(v1.z, v2.z, v3.z); \
+    r.w = fnc(v1.w, v2.w, v3.w); \
+    return r; \
+}
+
+#define FN_FUNC_FN_FN_PIN(fnc) \
+_RS_STATIC float2 __attribute__((overloadable)) fnc(float2 v1, float2 v2, int2 *v3) { \
+    float2 r; \
+    int q; \
+    r.x = fnc(v1.x, v2.x, &q); \
+    v3->x = q; \
+    r.y = fnc(v1.y, v2.y, &q); \
+    v3->y = q; \
+    return r; \
+} \
+_RS_STATIC float3 __attribute__((overloadable)) fnc(float3 v1, float3 v2, int3 *v3) { \
+    float3 r; \
+    int q; \
+    r.x = fnc(v1.x, v2.x, &q); \
+    v3->x = q; \
+    r.y = fnc(v1.y, v2.y, &q); \
+    v3->y = q; \
+    r.z = fnc(v1.z, v2.z, &q); \
+    v3->z = q; \
+    return r; \
+} \
+_RS_STATIC float4 __attribute__((overloadable)) fnc(float4 v1, float4 v2, int4 *v3) { \
+    float4 r; \
+    int q; \
+    r.x = fnc(v1.x, v2.x, &q); \
+    v3->x = q; \
+    r.y = fnc(v1.y, v2.y, &q); \
+    v3->y = q; \
+    r.z = fnc(v1.z, v2.z, &q); \
+    v3->z = q; \
+    r.w = fnc(v1.w, v2.w, &q); \
+    v3->w = q; \
+    return r; \
+}
+
+
+
 extern float __attribute__((overloadable)) acos(float);
-DEF_FUNC_1(acos)
+FN_FUNC_FN(acos)
 
 extern float __attribute__((overloadable)) acosh(float);
-DEF_FUNC_1(acosh)
+FN_FUNC_FN(acosh)
 
 _RS_STATIC float __attribute__((overloadable)) acospi(float v) {
     return acos(v) / M_PI;
 }
-DEF_FUNC_1(acospi)
+FN_FUNC_FN(acospi)
 
 extern float __attribute__((overloadable)) asin(float);
-DEF_FUNC_1(asin)
+FN_FUNC_FN(asin)
 
 extern float __attribute__((overloadable)) asinh(float);
-DEF_FUNC_1(asinh)
+FN_FUNC_FN(asinh)
 
 _RS_STATIC float __attribute__((overloadable)) asinpi(float v) {
     return asin(v) / M_PI;
 }
-DEF_FUNC_1(asinpi)
+FN_FUNC_FN(asinpi)
 
 extern float __attribute__((overloadable)) atan(float);
-DEF_FUNC_1(atan)
+FN_FUNC_FN(atan)
 
 extern float __attribute__((overloadable)) atan2(float, float);
-DEF_FUNC_2(atan2)
+FN_FUNC_FN_FN(atan2)
 
 extern float __attribute__((overloadable)) atanh(float);
-DEF_FUNC_1(atanh)
+FN_FUNC_FN(atanh)
 
 _RS_STATIC float __attribute__((overloadable)) atanpi(float v) {
     return atan(v) / M_PI;
 }
-DEF_FUNC_1(atanpi)
+FN_FUNC_FN(atanpi)
 
 _RS_STATIC float __attribute__((overloadable)) atan2pi(float y, float x) {
     return atan2(y, x) / M_PI;
 }
-DEF_FUNC_2(atan2pi)
+FN_FUNC_FN_FN(atan2pi)
 
 extern float __attribute__((overloadable)) cbrt(float);
-DEF_FUNC_1(cbrt)
+FN_FUNC_FN(cbrt)
 
 extern float __attribute__((overloadable)) ceil(float);
-DEF_FUNC_1(ceil)
+FN_FUNC_FN(ceil)
 
 extern float __attribute__((overloadable)) copysign(float, float);
-DEF_FUNC_2(copysign)
+FN_FUNC_FN_FN(copysign)
 
 extern float __attribute__((overloadable)) cos(float);
-DEF_FUNC_1(cos)
+FN_FUNC_FN(cos)
 
 extern float __attribute__((overloadable)) cosh(float);
-DEF_FUNC_1(cosh)
+FN_FUNC_FN(cosh)
 
 _RS_STATIC float __attribute__((overloadable)) cospi(float v) {
     return cos(v * M_PI);
 }
-DEF_FUNC_1(cospi)
+FN_FUNC_FN(cospi)
 
 extern float __attribute__((overloadable)) erfc(float);
-DEF_FUNC_1(erfc)
+FN_FUNC_FN(erfc)
 
 extern float __attribute__((overloadable)) erf(float);
-DEF_FUNC_1(erf)
+FN_FUNC_FN(erf)
 
 extern float __attribute__((overloadable)) exp(float);
-DEF_FUNC_1(exp)
+FN_FUNC_FN(exp)
 
 extern float __attribute__((overloadable)) exp2(float);
-DEF_FUNC_1(exp2)
+FN_FUNC_FN(exp2)
 
 extern float __attribute__((overloadable)) pow(float, float);
 _RS_STATIC float __attribute__((overloadable)) exp10(float v) {
     return pow(10.f, v);
 }
-DEF_FUNC_1(exp10)
+FN_FUNC_FN(exp10)
 
 extern float __attribute__((overloadable)) expm1(float);
-DEF_FUNC_1(expm1)
+FN_FUNC_FN(expm1)
 
 extern float __attribute__((overloadable)) fabs(float);
-DEF_FUNC_1(fabs)
+FN_FUNC_FN(fabs)
 
 extern float __attribute__((overloadable)) fdim(float, float);
-DEF_FUNC_2(fdim)
+FN_FUNC_FN_FN(fdim)
 
 extern float __attribute__((overloadable)) floor(float);
-DEF_FUNC_1(floor)
+FN_FUNC_FN(floor)
 
 extern float __attribute__((overloadable)) fma(float, float, float);
-extern float2 __attribute__((overloadable)) fma(float2, float2, float2);
+FN_FUNC_FN_FN_FN(fma)
-extern float3 __attribute__((overloadable)) fma(float3, float3, float3);
-extern float4 __attribute__((overloadable)) fma(float4, float4, float4);
 
 extern float __attribute__((overloadable)) fmax(float, float);
-DEF_FUNC_2(fmax);
+FN_FUNC_FN_FN(fmax);
-DEF_FUNC_2F(fmax);
+FN_FUNC_FN_F(fmax);
 
 extern float __attribute__((overloadable)) fmin(float, float);
-DEF_FUNC_2(fmin);
+FN_FUNC_FN_FN(fmin);
-DEF_FUNC_2F(fmin);
+FN_FUNC_FN_F(fmin);
 
 extern float __attribute__((overloadable)) fmod(float, float);
-DEF_FUNC_2(fmod)
+FN_FUNC_FN_FN(fmod)
 
 _RS_STATIC float __attribute__((overloadable)) fract(float v, float *iptr) {
     int i = (int)floor(v);
@@ -315,64 +454,54 @@ _RS_STATIC float4 __attribute__((overloadable)) fract(float4 v, float4 *iptr) {
     return r;
 }
 
-extern float __attribute__((overloadable)) frexp(float, float *);
+extern float __attribute__((overloadable)) frexp(float, int *);
-extern float2 __attribute__((overloadable)) frexp(float2, float2 *);
+FN_FUNC_FN_PIN(frexp)
-extern float3 __attribute__((overloadable)) frexp(float3, float3 *);
-extern float4 __attribute__((overloadable)) frexp(float4, float4 *);
 
 extern float __attribute__((overloadable)) hypot(float, float);
-DEF_FUNC_2(hypot)
+FN_FUNC_FN_FN(hypot)
 
 extern int __attribute__((overloadable)) ilogb(float);
-DEF_FUNC_1_RI(ilogb)
+IN_FUNC_FN(ilogb)
 
 extern float __attribute__((overloadable)) ldexp(float, int);
-extern float2 __attribute__((overloadable)) ldexp(float2, int2);
+FN_FUNC_FN_IN(ldexp)
-extern float3 __attribute__((overloadable)) ldexp(float3, int3);
+FN_FUNC_FN_I(ldexp)
-extern float4 __attribute__((overloadable)) ldexp(float4, int4);
-extern float2 __attribute__((overloadable)) ldexp(float2, int);
-extern float3 __attribute__((overloadable)) ldexp(float3, int);
-extern float4 __attribute__((overloadable)) ldexp(float4, int);
 
 extern float __attribute__((overloadable)) lgamma(float);
-DEF_FUNC_1(lgamma)
+FN_FUNC_FN(lgamma)
-extern float __attribute__((overloadable)) lgamma(float, float *);
+extern float __attribute__((overloadable)) lgamma(float, int*);
-extern float2 __attribute__((overloadable)) lgamma(float2, float2 *);
+FN_FUNC_FN_PIN(lgamma)
-extern float3 __attribute__((overloadable)) lgamma(float3, float3 *);
-extern float4 __attribute__((overloadable)) lgamma(float4, float4 *);
 
 extern float __attribute__((overloadable)) log(float);
-DEF_FUNC_1(log)
+FN_FUNC_FN(log)
 
 
 extern float __attribute__((overloadable)) log10(float);
-DEF_FUNC_1(log10)
+FN_FUNC_FN(log10)
 
 _RS_STATIC float __attribute__((overloadable)) log2(float v) {
     return log10(v) / log10(2.f);
 }
-DEF_FUNC_1(log2)
+FN_FUNC_FN(log2)
 
 extern float __attribute__((overloadable)) log1p(float);
-DEF_FUNC_1(log1p)
+FN_FUNC_FN(log1p)
 
 extern float __attribute__((overloadable)) logb(float);
-DEF_FUNC_1(logb)
+FN_FUNC_FN(logb)
 
 extern float __attribute__((overloadable)) mad(float, float, float);
-extern float2 __attribute__((overloadable)) mad(float2, float2, float2);
+FN_FUNC_FN_FN_FN(mad)
-extern float3 __attribute__((overloadable)) mad(float3, float3, float3);
-extern float4 __attribute__((overloadable)) mad(float4, float4, float4);
 
 extern float __attribute__((overloadable)) modf(float, float *);
-DEF_FUNC_2P(modf);
+FN_FUNC_FN_PFN(modf);
 
 //extern float __attribute__((overloadable)) nan(uint);
 
 extern float __attribute__((overloadable)) nextafter(float, float);
-DEF_FUNC_2(nextafter)
+FN_FUNC_FN_FN(nextafter)
 
-DEF_FUNC_2(pow)
+FN_FUNC_FN_FN(pow)
 
 _RS_STATIC float __attribute__((overloadable)) pown(float v, int p) {
     return pow(v, (float)p);
@@ -401,15 +530,13 @@ _RS_STATIC float4 __attribute__((overloadable)) powr(float4 v, float4 p) {
 }
 
 extern float __attribute__((overloadable)) remainder(float, float);
-DEF_FUNC_2(remainder)
+FN_FUNC_FN_FN(remainder)
 
 extern float __attribute__((overloadable)) remquo(float, float, int *);
-extern float2 __attribute__((overloadable)) remquo(float2, float2, int2 *);
+FN_FUNC_FN_FN_PIN(remquo)
-extern float3 __attribute__((overloadable)) remquo(float3, float3, int3 *);
-extern float4 __attribute__((overloadable)) remquo(float4, float4, int4 *);
 
 extern float __attribute__((overloadable)) rint(float);
-DEF_FUNC_1(rint)
+FN_FUNC_FN(rint)
 
 _RS_STATIC float __attribute__((overloadable)) rootn(float v, int r) {
     return pow(v, 1.f / r);
@@ -428,16 +555,16 @@ _RS_STATIC float4 __attribute__((overloadable)) rootn(float4 v, int4 r) {
 }
 
 extern float __attribute__((overloadable)) round(float);
-DEF_FUNC_1(round)
+FN_FUNC_FN(round)
 
 extern float __attribute__((overloadable)) sqrt(float);
 _RS_STATIC float __attribute__((overloadable)) rsqrt(float v) {
     return 1.f / sqrt(v);
 }
-DEF_FUNC_1(rsqrt)
+FN_FUNC_FN(rsqrt)
 
 extern float __attribute__((overloadable)) sin(float);
-DEF_FUNC_1(sin)
+FN_FUNC_FN(sin)
 
 _RS_STATIC float __attribute__((overloadable)) sincos(float v, float *cosptr) {
     *cosptr = cos(v);
@@ -457,35 +584,35 @@ _RS_STATIC float4 __attribute__((overloadable)) sincos(float4 v, float4 *cosptr)
 }
 
 extern float __attribute__((overloadable)) sinh(float);
-DEF_FUNC_1(sinh)
+FN_FUNC_FN(sinh)
 
 _RS_STATIC float __attribute__((overloadable)) sinpi(float v) {
     return sin(v * M_PI);
 }
-DEF_FUNC_1(sinpi)
+FN_FUNC_FN(sinpi)
 
-DEF_FUNC_1(sqrt)
+FN_FUNC_FN(sqrt)
 
 extern float __attribute__((overloadable)) tan(float);
-DEF_FUNC_1(tan)
+FN_FUNC_FN(tan)
 
 extern float __attribute__((overloadable)) tanh(float);
-DEF_FUNC_1(tanh)
+FN_FUNC_FN(tanh)
 
 _RS_STATIC float __attribute__((overloadable)) tanpi(float v) {
     return tan(v * M_PI);
 }
-DEF_FUNC_1(tanpi)
+FN_FUNC_FN(tanpi)
 
 extern float __attribute__((overloadable)) tgamma(float);
-DEF_FUNC_1(tgamma)
+FN_FUNC_FN(tgamma)
 
 extern float __attribute__((overloadable)) trunc(float);
-DEF_FUNC_1(trunc)
+FN_FUNC_FN(trunc)
 
 // Int ops (partial), 6.11.3
 
-#define DEF_RIFUNC_1(typeout, typein, fnc)                          \
+#define XN_FUNC_YN(typeout, fnc, typein)                            \
 extern typeout __attribute__((overloadable)) fnc(typein);           \
 _RS_STATIC typeout##2 __attribute__((overloadable)) fnc(typein##2 v) {  \
     typeout##2 r;                                                   \
@@ -509,20 +636,20 @@ _RS_STATIC typeout##4 __attribute__((overloadable)) fnc(typein##4 v) {  \
     return r;                                                       \
 }
 
-#define DEF_UIFUNC_1(fnc)           \
+#define UIN_FUNC_IN(fnc)           \
-DEF_RIFUNC_1(uchar, char, fnc)      \
+XN_FUNC_YN(uchar, fnc, char)      \
-DEF_RIFUNC_1(ushort, short, fnc)    \
+XN_FUNC_YN(ushort, fnc, short)    \
-DEF_RIFUNC_1(uint, int, fnc)
+XN_FUNC_YN(uint, fnc, int)
 
-#define DEF_IFUNC_1(fnc)            \
+#define IN_FUNC_IN(fnc)            \
-DEF_RIFUNC_1(uchar, uchar, fnc)     \
+XN_FUNC_YN(uchar, fnc, uchar)     \
-DEF_RIFUNC_1(char, char, fnc)       \
+XN_FUNC_YN(char, fnc, char)       \
-DEF_RIFUNC_1(ushort, ushort, fnc)   \
+XN_FUNC_YN(ushort, fnc, ushort)   \
-DEF_RIFUNC_1(short, short, fnc)     \
+XN_FUNC_YN(short, fnc, short)     \
-DEF_RIFUNC_1(uint, uint, fnc)       \
+XN_FUNC_YN(uint, fnc, uint)       \
-DEF_RIFUNC_1(int, int, fnc)
+XN_FUNC_YN(int, fnc, int)
 
-#define DEF_RIFUNC_2(type, fnc, body)                                       \
+#define XN_FUNC_XN_XN_BODY(type, fnc, body)                                       \
 _RS_STATIC type __attribute__((overloadable)) fnc(type v1, type v2) {           \
     return body;                                                            \
 }                                                                           \
@@ -548,23 +675,23 @@ _RS_STATIC type##4 __attribute__((overloadable)) fnc(type##4 v1, type##4 v2) {
     return r;                                                               \
 }                                                                           \
 
-#define DEF_IFUNC_2(fnc, body)  \
+#define IN_FUNC_IN_IN_BODY(fnc, body)  \
-DEF_RIFUNC_2(uchar, fnc, body)  \
+XN_FUNC_XN_XN_BODY(uchar, fnc, body)  \
-DEF_RIFUNC_2(char, fnc, body)   \
+XN_FUNC_XN_XN_BODY(char, fnc, body)   \
-DEF_RIFUNC_2(ushort, fnc, body) \
+XN_FUNC_XN_XN_BODY(ushort, fnc, body) \
-DEF_RIFUNC_2(short, fnc, body)  \
+XN_FUNC_XN_XN_BODY(short, fnc, body)  \
-DEF_RIFUNC_2(uint, fnc, body)   \
+XN_FUNC_XN_XN_BODY(uint, fnc, body)   \
-DEF_RIFUNC_2(int, fnc, body)    \
+XN_FUNC_XN_XN_BODY(int, fnc, body)    \
-DEF_RIFUNC_2(float, fnc, body)
+XN_FUNC_XN_XN_BODY(float, fnc, body)
 
-DEF_UIFUNC_1(abs)
+UIN_FUNC_IN(abs)
-DEF_IFUNC_1(clz)
+IN_FUNC_IN(clz)
 
-DEF_IFUNC_2(min, (v1 < v2 ? v1 : v2))
+IN_FUNC_IN_IN_BODY(min, (v1 < v2 ? v1 : v2))
-DEF_FUNC_2F(min)
+FN_FUNC_FN_F(min)
 
-DEF_IFUNC_2(max, (v1 > v2 ? v1 : v2))
+IN_FUNC_IN_IN_BODY(max, (v1 > v2 ? v1 : v2))
-DEF_FUNC_2F(max)
+FN_FUNC_FN_F(max)
 
 // 6.11.4
 
@@ -617,7 +744,7 @@ _RS_STATIC float4 __attribute__((overloadable)) clamp(float4 amount, float low,
 _RS_STATIC float __attribute__((overloadable)) degrees(float radians) {
     return radians * (180.f / M_PI);
 }
-DEF_FUNC_1(degrees)
+FN_FUNC_FN(degrees)
 
 _RS_STATIC float __attribute__((overloadable)) mix(float start, float stop, float amount) {
     return start + (stop - start) * amount;
@@ -644,7 +771,7 @@ _RS_STATIC float4 __attribute__((overloadable)) mix(float4 start, float4 stop, f
 _RS_STATIC float __attribute__((overloadable)) radians(float degrees) {
     return degrees * (M_PI / 180.f);
 }
-DEF_FUNC_1(radians)
+FN_FUNC_FN(radians)
 
 _RS_STATIC float __attribute__((overloadable)) step(float edge, float v) {
     return (v < edge) ? 0.f : 1.f;
@@ -705,7 +832,7 @@ _RS_STATIC float __attribute__((overloadable)) sign(float v) {
     if (v < 0) return -1.f;
     return v;
 }
-DEF_FUNC_1(sign)
+FN_FUNC_FN(sign)
 
 // 6.11.5
 _RS_STATIC float3 __attribute__((overloadable)) cross(float3 lhs, float3 rhs) {
@@ -779,15 +906,21 @@ _RS_STATIC float4 __attribute__((overloadable)) normalize(float4 v) {
 
 #undef CVT_FUNC
 #undef CVT_FUNC_2
-#undef DEF_FUNC_1
+#undef FN_FUNC_FN
-#undef DEF_FUNC_1_RI
+#undef IN_FUNC_FN
-#undef DEF_FUNC_2
+#undef FN_FUNC_FN_FN
-#undef DEF_FUNC_2F
+#undef FN_FUNC_FN_F
-#undef DEF_RIFUNC_1
+#undef FN_FUNC_FN_IN
-#undef DEF_UIFUNC_1
+#undef FN_FUNC_FN_I
-#undef DEF_IFUNC_1
+#undef FN_FUNC_FN_PFN
-#undef DEF_RIFUNC_2
+#undef FN_FUNC_FN_PIN
-#undef DEF_IFUNC_2
+#undef FN_FUNC_FN_FN_FN
+#undef FN_FUNC_FN_FN_PIN
+#undef XN_FUNC_YN
+#undef UIN_FUNC_IN
+#undef IN_FUNC_IN
+#undef XN_FUNC_XN_XN_BODY
+#undef IN_FUNC_IN_IN_BODY
 #undef _RS_STATIC
 
 #endif