diff --git a/.gitignore b/.gitignore
index d03b6fb..695d9d9 100644
--- a/.gitignore
+++ b/.gitignore
@@ -18,6 +18,8 @@ Debug/*
 
 src/bindgen_core_api.c
 src/bindgen_gles_api.c
-*bind_gen.c
 sdk/io_stubs.c
-sdk/orca_surface.c
\ No newline at end of file
+sdk/orca_surface.c
+*bind_gen.c
+
+.vscode/settings.json
diff --git a/build.bat b/build.bat
index 92125c1..a1cd857 100644
--- a/build.bat
+++ b/build.bat
@@ -13,7 +13,7 @@ if %target% == wasm3 (
 	set wasm3_sources=/I .\ext\wasm3\source\*.c
 
 	for %%f in ( .\ext\wasm3\source\*.c ) do (
-		cl /nologo /Zi /Zc:preprocessor /c /Fo:bin\obj\%%~nf.obj %wasm3_includes% %%f
+		cl /nologo /Zi /Zc:preprocessor /O2 /c /Fo:bin\obj\%%~nf.obj %wasm3_includes% %%f
 	)
 	lib /nologo /out:bin\wasm3.lib bin\obj\*.obj
 )
diff --git a/cstdlib/include/math.h b/cstdlib/include/math.h
index 92a168e..58c2697 100644
--- a/cstdlib/include/math.h
+++ b/cstdlib/include/math.h
@@ -54,6 +54,9 @@ double      acos(double);
 
 double      ceil(double);
 
+double      cos(double);
+float       cosf(float);
+
 double      fabs(double);
 
 double      floor(double);
@@ -62,7 +65,27 @@ double      fmod(double, double);
 
 double      pow(double, double);
 
+double      scalbn(double, int);
+
+double      sin(double);
+float       sinf(float);
+
 double      sqrt(double);
+float       sqrtf(float);
+
+#define M_E             2.7182818284590452354   /* e */
+#define M_LOG2E         1.4426950408889634074   /* log_2 e */
+#define M_LOG10E        0.43429448190325182765  /* log_10 e */
+#define M_LN2           0.69314718055994530942  /* log_e 2 */
+#define M_LN10          2.30258509299404568402  /* log_e 10 */
+#define M_PI            3.14159265358979323846  /* pi */
+#define M_PI_2          1.57079632679489661923  /* pi/2 */
+#define M_PI_4          0.78539816339744830962  /* pi/4 */
+#define M_1_PI          0.31830988618379067154  /* 1/pi */
+#define M_2_PI          0.63661977236758134308  /* 2/pi */
+#define M_2_SQRTPI      1.12837916709551257390  /* 2/sqrt(pi) */
+#define M_SQRT2         1.41421356237309504880  /* sqrt(2) */
+#define M_SQRT1_2       0.70710678118654752440  /* 1/sqrt(2) */
 
 #ifdef __cplusplus
 }
diff --git a/cstdlib/src/__cos.c b/cstdlib/src/__cos.c
new file mode 100644
index 0000000..46cefb3
--- /dev/null
+++ b/cstdlib/src/__cos.c
@@ -0,0 +1,71 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/k_cos.c */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunSoft, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+/*
+ * __cos( x,  y )
+ * kernel cos function on [-pi/4, pi/4], pi/4 ~ 0.785398164
+ * Input x is assumed to be bounded by ~pi/4 in magnitude.
+ * Input y is the tail of x.
+ *
+ * Algorithm
+ *      1. Since cos(-x) = cos(x), we need only to consider positive x.
+ *      2. if x < 2^-27 (hx<0x3e400000 0), return 1 with inexact if x!=0.
+ *      3. cos(x) is approximated by a polynomial of degree 14 on
+ *         [0,pi/4]
+ *                                       4            14
+ *              cos(x) ~ 1 - x*x/2 + C1*x + ... + C6*x
+ *         where the remez error is
+ *
+ *      |              2     4     6     8     10    12     14 |     -58
+ *      |cos(x)-(1-.5*x +C1*x +C2*x +C3*x +C4*x +C5*x  +C6*x  )| <= 2
+ *      |                                                      |
+ *
+ *                     4     6     8     10    12     14
+ *      4. let r = C1*x +C2*x +C3*x +C4*x +C5*x  +C6*x  , then
+ *             cos(x) ~ 1 - x*x/2 + r
+ *         since cos(x+y) ~ cos(x) - sin(x)*y
+ *                        ~ cos(x) - x*y,
+ *         a correction term is necessary in cos(x) and hence
+ *              cos(x+y) = 1 - (x*x/2 - (r - x*y))
+ *         For better accuracy, rearrange to
+ *              cos(x+y) ~ w + (tmp + (r-x*y))
+ *         where w = 1 - x*x/2 and tmp is a tiny correction term
+ *         (1 - x*x/2 == w + tmp exactly in infinite precision).
+ *         The exactness of w + tmp in infinite precision depends on w
+ *         and tmp having the same precision as x.  If they have extra
+ *         precision due to compiler bugs, then the extra precision is
+ *         only good provided it is retained in all terms of the final
+ *         expression for cos().  Retention happens in all cases tested
+ *         under FreeBSD, so don't pessimize things by forcibly clipping
+ *         any extra precision in w.
+ */
+
+#include "libm.h"
+
+static const double
+C1  =  4.16666666666666019037e-02, /* 0x3FA55555, 0x5555554C */
+C2  = -1.38888888888741095749e-03, /* 0xBF56C16C, 0x16C15177 */
+C3  =  2.48015872894767294178e-05, /* 0x3EFA01A0, 0x19CB1590 */
+C4  = -2.75573143513906633035e-07, /* 0xBE927E4F, 0x809C52AD */
+C5  =  2.08757232129817482790e-09, /* 0x3E21EE9E, 0xBDB4B1C4 */
+C6  = -1.13596475577881948265e-11; /* 0xBDA8FAE9, 0xBE8838D4 */
+
+double __cos(double x, double y)
+{
+	double_t hz,z,r,w;
+
+	z  = x*x;
+	w  = z*z;
+	r  = z*(C1+z*(C2+z*C3)) + w*w*(C4+z*(C5+z*C6));
+	hz = 0.5*z;
+	w  = 1.0-hz;
+	return w + (((1.0-w)-hz) + (z*r-x*y));
+}
diff --git a/cstdlib/src/__cosdf.c b/cstdlib/src/__cosdf.c
new file mode 100644
index 0000000..2124989
--- /dev/null
+++ b/cstdlib/src/__cosdf.c
@@ -0,0 +1,35 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/k_cosf.c */
+/*
+ * Conversion to float by Ian Lance Taylor, Cygnus Support, ian@cygnus.com.
+ * Debugged and optimized by Bruce D. Evans.
+ */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+
+#include "libm.h"
+
+/* |cos(x) - c(x)| < 2**-34.1 (~[-5.37e-11, 5.295e-11]). */
+static const double
+C0  = -0x1ffffffd0c5e81.0p-54, /* -0.499999997251031003120 */
+C1  =  0x155553e1053a42.0p-57, /*  0.0416666233237390631894 */
+C2  = -0x16c087e80f1e27.0p-62, /* -0.00138867637746099294692 */
+C3  =  0x199342e0ee5069.0p-68; /*  0.0000243904487962774090654 */
+
+float __cosdf(double x)
+{
+	double_t r, w, z;
+
+	/* Try to optimize for parallel evaluation as in __tandf.c. */
+	z = x*x;
+	w = z*z;
+	r = C2+z*C3;
+	return ((1.0+z*C0) + w*C1) + (w*z)*r;
+}
diff --git a/cstdlib/src/__math_invalidf.c b/cstdlib/src/__math_invalidf.c
new file mode 100644
index 0000000..357d4b1
--- /dev/null
+++ b/cstdlib/src/__math_invalidf.c
@@ -0,0 +1,6 @@
+#include "libm.h"
+
+float __math_invalidf(float x)
+{
+	return (x - x) / (x - x);
+}
diff --git a/cstdlib/src/__rem_pio2.c b/cstdlib/src/__rem_pio2.c
new file mode 100644
index 0000000..dcf672f
--- /dev/null
+++ b/cstdlib/src/__rem_pio2.c
@@ -0,0 +1,190 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/e_rem_pio2.c */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunSoft, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ *
+ * Optimized by Bruce D. Evans.
+ */
+/* __rem_pio2(x,y)
+ *
+ * return the remainder of x rem pi/2 in y[0]+y[1]
+ * use __rem_pio2_large() for large x
+ */
+
+#include "libm.h"
+
+#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+
+/*
+ * invpio2:  53 bits of 2/pi
+ * pio2_1:   first  33 bit of pi/2
+ * pio2_1t:  pi/2 - pio2_1
+ * pio2_2:   second 33 bit of pi/2
+ * pio2_2t:  pi/2 - (pio2_1+pio2_2)
+ * pio2_3:   third  33 bit of pi/2
+ * pio2_3t:  pi/2 - (pio2_1+pio2_2+pio2_3)
+ */
+static const double
+toint   = 1.5/EPS,
+pio4    = 0x1.921fb54442d18p-1,
+invpio2 = 6.36619772367581382433e-01, /* 0x3FE45F30, 0x6DC9C883 */
+pio2_1  = 1.57079632673412561417e+00, /* 0x3FF921FB, 0x54400000 */
+pio2_1t = 6.07710050650619224932e-11, /* 0x3DD0B461, 0x1A626331 */
+pio2_2  = 6.07710050630396597660e-11, /* 0x3DD0B461, 0x1A600000 */
+pio2_2t = 2.02226624879595063154e-21, /* 0x3BA3198A, 0x2E037073 */
+pio2_3  = 2.02226624871116645580e-21, /* 0x3BA3198A, 0x2E000000 */
+pio2_3t = 8.47842766036889956997e-32; /* 0x397B839A, 0x252049C1 */
+
+/* caller must handle the case when reduction is not needed: |x| ~<= pi/4 */
+int __rem_pio2(double x, double *y)
+{
+	union {double f; uint64_t i;} u = {x};
+	double_t z,w,t,r,fn;
+	double tx[3],ty[2];
+	uint32_t ix;
+	int sign, n, ex, ey, i;
+
+	sign = u.i>>63;
+	ix = u.i>>32 & 0x7fffffff;
+	if (ix <= 0x400f6a7a) {  /* |x| ~<= 5pi/4 */
+		if ((ix & 0xfffff) == 0x921fb)  /* |x| ~= pi/2 or 2pi/2 */
+			goto medium;  /* cancellation -- use medium case */
+		if (ix <= 0x4002d97c) {  /* |x| ~<= 3pi/4 */
+			if (!sign) {
+				z = x - pio2_1;  /* one round good to 85 bits */
+				y[0] = z - pio2_1t;
+				y[1] = (z-y[0]) - pio2_1t;
+				return 1;
+			} else {
+				z = x + pio2_1;
+				y[0] = z + pio2_1t;
+				y[1] = (z-y[0]) + pio2_1t;
+				return -1;
+			}
+		} else {
+			if (!sign) {
+				z = x - 2*pio2_1;
+				y[0] = z - 2*pio2_1t;
+				y[1] = (z-y[0]) - 2*pio2_1t;
+				return 2;
+			} else {
+				z = x + 2*pio2_1;
+				y[0] = z + 2*pio2_1t;
+				y[1] = (z-y[0]) + 2*pio2_1t;
+				return -2;
+			}
+		}
+	}
+	if (ix <= 0x401c463b) {  /* |x| ~<= 9pi/4 */
+		if (ix <= 0x4015fdbc) {  /* |x| ~<= 7pi/4 */
+			if (ix == 0x4012d97c)  /* |x| ~= 3pi/2 */
+				goto medium;
+			if (!sign) {
+				z = x - 3*pio2_1;
+				y[0] = z - 3*pio2_1t;
+				y[1] = (z-y[0]) - 3*pio2_1t;
+				return 3;
+			} else {
+				z = x + 3*pio2_1;
+				y[0] = z + 3*pio2_1t;
+				y[1] = (z-y[0]) + 3*pio2_1t;
+				return -3;
+			}
+		} else {
+			if (ix == 0x401921fb)  /* |x| ~= 4pi/2 */
+				goto medium;
+			if (!sign) {
+				z = x - 4*pio2_1;
+				y[0] = z - 4*pio2_1t;
+				y[1] = (z-y[0]) - 4*pio2_1t;
+				return 4;
+			} else {
+				z = x + 4*pio2_1;
+				y[0] = z + 4*pio2_1t;
+				y[1] = (z-y[0]) + 4*pio2_1t;
+				return -4;
+			}
+		}
+	}
+	if (ix < 0x413921fb) {  /* |x| ~< 2^20*(pi/2), medium size */
+medium:
+		/* rint(x/(pi/2)) */
+		fn = (double_t)x*invpio2 + toint - toint;
+		n = (int32_t)fn;
+		r = x - fn*pio2_1;
+		w = fn*pio2_1t;  /* 1st round, good to 85 bits */
+		/* Matters with directed rounding. */
+		if (predict_false(r - w < -pio4)) {
+			n--;
+			fn--;
+			r = x - fn*pio2_1;
+			w = fn*pio2_1t;
+		} else if (predict_false(r - w > pio4)) {
+			n++;
+			fn++;
+			r = x - fn*pio2_1;
+			w = fn*pio2_1t;
+		}
+		y[0] = r - w;
+		u.f = y[0];
+		ey = u.i>>52 & 0x7ff;
+		ex = ix>>20;
+		if (ex - ey > 16) { /* 2nd round, good to 118 bits */
+			t = r;
+			w = fn*pio2_2;
+			r = t - w;
+			w = fn*pio2_2t - ((t-r)-w);
+			y[0] = r - w;
+			u.f = y[0];
+			ey = u.i>>52 & 0x7ff;
+			if (ex - ey > 49) {  /* 3rd round, good to 151 bits, covers all cases */
+				t = r;
+				w = fn*pio2_3;
+				r = t - w;
+				w = fn*pio2_3t - ((t-r)-w);
+				y[0] = r - w;
+			}
+		}
+		y[1] = (r - y[0]) - w;
+		return n;
+	}
+	/*
+	 * all other (large) arguments
+	 */
+	if (ix >= 0x7ff00000) {  /* x is inf or NaN */
+		y[0] = y[1] = x - x;
+		return 0;
+	}
+	/* set z = scalbn(|x|,-ilogb(x)+23) */
+	u.f = x;
+	u.i &= (uint64_t)-1>>12;
+	u.i |= (uint64_t)(0x3ff + 23)<<52;
+	z = u.f;
+	for (i=0; i < 2; i++) {
+		tx[i] = (double)(int32_t)z;
+		z     = (z-tx[i])*0x1p24;
+	}
+	tx[i] = z;
+	/* skip zero terms, first term is non-zero */
+	while (tx[i] == 0.0)
+		i--;
+	n = __rem_pio2_large(tx,ty,(int)(ix>>20)-(0x3ff+23),i+1,1);
+	if (sign) {
+		y[0] = -ty[0];
+		y[1] = -ty[1];
+		return -n;
+	}
+	y[0] = ty[0];
+	y[1] = ty[1];
+	return n;
+}
diff --git a/cstdlib/src/__rem_pio2_large.c b/cstdlib/src/__rem_pio2_large.c
new file mode 100644
index 0000000..958f28c
--- /dev/null
+++ b/cstdlib/src/__rem_pio2_large.c
@@ -0,0 +1,442 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/k_rem_pio2.c */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunSoft, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+/*
+ * __rem_pio2_large(x,y,e0,nx,prec)
+ * double x[],y[]; int e0,nx,prec;
+ *
+ * __rem_pio2_large return the last three digits of N with
+ *              y = x - N*pi/2
+ * so that |y| < pi/2.
+ *
+ * The method is to compute the integer (mod 8) and fraction parts of
+ * (2/pi)*x without doing the full multiplication. In general we
+ * skip the part of the product that are known to be a huge integer (
+ * more accurately, = 0 mod 8 ). Thus the number of operations are
+ * independent of the exponent of the input.
+ *
+ * (2/pi) is represented by an array of 24-bit integers in ipio2[].
+ *
+ * Input parameters:
+ *      x[]     The input value (must be positive) is broken into nx
+ *              pieces of 24-bit integers in double precision format.
+ *              x[i] will be the i-th 24 bit of x. The scaled exponent
+ *              of x[0] is given in input parameter e0 (i.e., x[0]*2^e0
+ *              match x's up to 24 bits.
+ *
+ *              Example of breaking a double positive z into x[0]+x[1]+x[2]:
+ *                      e0 = ilogb(z)-23
+ *                      z  = scalbn(z,-e0)
+ *              for i = 0,1,2
+ *                      x[i] = floor(z)
+ *                      z    = (z-x[i])*2**24
+ *
+ *
+ *      y[]     ouput result in an array of double precision numbers.
+ *              The dimension of y[] is:
+ *                      24-bit  precision       1
+ *                      53-bit  precision       2
+ *                      64-bit  precision       2
+ *                      113-bit precision       3
+ *              The actual value is the sum of them. Thus for 113-bit
+ *              precison, one may have to do something like:
+ *
+ *              long double t,w,r_head, r_tail;
+ *              t = (long double)y[2] + (long double)y[1];
+ *              w = (long double)y[0];
+ *              r_head = t+w;
+ *              r_tail = w - (r_head - t);
+ *
+ *      e0      The exponent of x[0]. Must be <= 16360 or you need to
+ *              expand the ipio2 table.
+ *
+ *      nx      dimension of x[]
+ *
+ *      prec    an integer indicating the precision:
+ *                      0       24  bits (single)
+ *                      1       53  bits (double)
+ *                      2       64  bits (extended)
+ *                      3       113 bits (quad)
+ *
+ * External function:
+ *      double scalbn(), floor();
+ *
+ *
+ * Here is the description of some local variables:
+ *
+ *      jk      jk+1 is the initial number of terms of ipio2[] needed
+ *              in the computation. The minimum and recommended value
+ *              for jk is 3,4,4,6 for single, double, extended, and quad.
+ *              jk+1 must be 2 larger than you might expect so that our
+ *              recomputation test works. (Up to 24 bits in the integer
+ *              part (the 24 bits of it that we compute) and 23 bits in
+ *              the fraction part may be lost to cancelation before we
+ *              recompute.)
+ *
+ *      jz      local integer variable indicating the number of
+ *              terms of ipio2[] used.
+ *
+ *      jx      nx - 1
+ *
+ *      jv      index for pointing to the suitable ipio2[] for the
+ *              computation. In general, we want
+ *                      ( 2^e0*x[0] * ipio2[jv-1]*2^(-24jv) )/8
+ *              is an integer. Thus
+ *                      e0-3-24*jv >= 0 or (e0-3)/24 >= jv
+ *              Hence jv = max(0,(e0-3)/24).
+ *
+ *      jp      jp+1 is the number of terms in PIo2[] needed, jp = jk.
+ *
+ *      q[]     double array with integral value, representing the
+ *              24-bits chunk of the product of x and 2/pi.
+ *
+ *      q0      the corresponding exponent of q[0]. Note that the
+ *              exponent for q[i] would be q0-24*i.
+ *
+ *      PIo2[]  double precision array, obtained by cutting pi/2
+ *              into 24 bits chunks.
+ *
+ *      f[]     ipio2[] in floating point
+ *
+ *      iq[]    integer array by breaking up q[] in 24-bits chunk.
+ *
+ *      fq[]    final product of x*(2/pi) in fq[0],..,fq[jk]
+ *
+ *      ih      integer. If >0 it indicates q[] is >= 0.5, hence
+ *              it also indicates the *sign* of the result.
+ *
+ */
+/*
+ * Constants:
+ * The hexadecimal values are the intended ones for the following
+ * constants. The decimal values may be used, provided that the
+ * compiler will convert from decimal to binary accurately enough
+ * to produce the hexadecimal values shown.
+ */
+
+#include "libm.h"
+
+static const int init_jk[] = {3,4,4,6}; /* initial value for jk */
+
+/*
+ * Table of constants for 2/pi, 396 Hex digits (476 decimal) of 2/pi
+ *
+ *              integer array, contains the (24*i)-th to (24*i+23)-th
+ *              bit of 2/pi after binary point. The corresponding
+ *              floating value is
+ *
+ *                      ipio2[i] * 2^(-24(i+1)).
+ *
+ * NB: This table must have at least (e0-3)/24 + jk terms.
+ *     For quad precision (e0 <= 16360, jk = 6), this is 686.
+ */
+static const int32_t ipio2[] = {
+0xA2F983, 0x6E4E44, 0x1529FC, 0x2757D1, 0xF534DD, 0xC0DB62,
+0x95993C, 0x439041, 0xFE5163, 0xABDEBB, 0xC561B7, 0x246E3A,
+0x424DD2, 0xE00649, 0x2EEA09, 0xD1921C, 0xFE1DEB, 0x1CB129,
+0xA73EE8, 0x8235F5, 0x2EBB44, 0x84E99C, 0x7026B4, 0x5F7E41,
+0x3991D6, 0x398353, 0x39F49C, 0x845F8B, 0xBDF928, 0x3B1FF8,
+0x97FFDE, 0x05980F, 0xEF2F11, 0x8B5A0A, 0x6D1F6D, 0x367ECF,
+0x27CB09, 0xB74F46, 0x3F669E, 0x5FEA2D, 0x7527BA, 0xC7EBE5,
+0xF17B3D, 0x0739F7, 0x8A5292, 0xEA6BFB, 0x5FB11F, 0x8D5D08,
+0x560330, 0x46FC7B, 0x6BABF0, 0xCFBC20, 0x9AF436, 0x1DA9E3,
+0x91615E, 0xE61B08, 0x659985, 0x5F14A0, 0x68408D, 0xFFD880,
+0x4D7327, 0x310606, 0x1556CA, 0x73A8C9, 0x60E27B, 0xC08C6B,
+
+#if LDBL_MAX_EXP > 1024
+0x47C419, 0xC367CD, 0xDCE809, 0x2A8359, 0xC4768B, 0x961CA6,
+0xDDAF44, 0xD15719, 0x053EA5, 0xFF0705, 0x3F7E33, 0xE832C2,
+0xDE4F98, 0x327DBB, 0xC33D26, 0xEF6B1E, 0x5EF89F, 0x3A1F35,
+0xCAF27F, 0x1D87F1, 0x21907C, 0x7C246A, 0xFA6ED5, 0x772D30,
+0x433B15, 0xC614B5, 0x9D19C3, 0xC2C4AD, 0x414D2C, 0x5D000C,
+0x467D86, 0x2D71E3, 0x9AC69B, 0x006233, 0x7CD2B4, 0x97A7B4,
+0xD55537, 0xF63ED7, 0x1810A3, 0xFC764D, 0x2A9D64, 0xABD770,
+0xF87C63, 0x57B07A, 0xE71517, 0x5649C0, 0xD9D63B, 0x3884A7,
+0xCB2324, 0x778AD6, 0x23545A, 0xB91F00, 0x1B0AF1, 0xDFCE19,
+0xFF319F, 0x6A1E66, 0x615799, 0x47FBAC, 0xD87F7E, 0xB76522,
+0x89E832, 0x60BFE6, 0xCDC4EF, 0x09366C, 0xD43F5D, 0xD7DE16,
+0xDE3B58, 0x929BDE, 0x2822D2, 0xE88628, 0x4D58E2, 0x32CAC6,
+0x16E308, 0xCB7DE0, 0x50C017, 0xA71DF3, 0x5BE018, 0x34132E,
+0x621283, 0x014883, 0x5B8EF5, 0x7FB0AD, 0xF2E91E, 0x434A48,
+0xD36710, 0xD8DDAA, 0x425FAE, 0xCE616A, 0xA4280A, 0xB499D3,
+0xF2A606, 0x7F775C, 0x83C2A3, 0x883C61, 0x78738A, 0x5A8CAF,
+0xBDD76F, 0x63A62D, 0xCBBFF4, 0xEF818D, 0x67C126, 0x45CA55,
+0x36D9CA, 0xD2A828, 0x8D61C2, 0x77C912, 0x142604, 0x9B4612,
+0xC459C4, 0x44C5C8, 0x91B24D, 0xF31700, 0xAD43D4, 0xE54929,
+0x10D5FD, 0xFCBE00, 0xCC941E, 0xEECE70, 0xF53E13, 0x80F1EC,
+0xC3E7B3, 0x28F8C7, 0x940593, 0x3E71C1, 0xB3092E, 0xF3450B,
+0x9C1288, 0x7B20AB, 0x9FB52E, 0xC29247, 0x2F327B, 0x6D550C,
+0x90A772, 0x1FE76B, 0x96CB31, 0x4A1679, 0xE27941, 0x89DFF4,
+0x9794E8, 0x84E6E2, 0x973199, 0x6BED88, 0x365F5F, 0x0EFDBB,
+0xB49A48, 0x6CA467, 0x427271, 0x325D8D, 0xB8159F, 0x09E5BC,
+0x25318D, 0x3974F7, 0x1C0530, 0x010C0D, 0x68084B, 0x58EE2C,
+0x90AA47, 0x02E774, 0x24D6BD, 0xA67DF7, 0x72486E, 0xEF169F,
+0xA6948E, 0xF691B4, 0x5153D1, 0xF20ACF, 0x339820, 0x7E4BF5,
+0x6863B2, 0x5F3EDD, 0x035D40, 0x7F8985, 0x295255, 0xC06437,
+0x10D86D, 0x324832, 0x754C5B, 0xD4714E, 0x6E5445, 0xC1090B,
+0x69F52A, 0xD56614, 0x9D0727, 0x50045D, 0xDB3BB4, 0xC576EA,
+0x17F987, 0x7D6B49, 0xBA271D, 0x296996, 0xACCCC6, 0x5414AD,
+0x6AE290, 0x89D988, 0x50722C, 0xBEA404, 0x940777, 0x7030F3,
+0x27FC00, 0xA871EA, 0x49C266, 0x3DE064, 0x83DD97, 0x973FA3,
+0xFD9443, 0x8C860D, 0xDE4131, 0x9D3992, 0x8C70DD, 0xE7B717,
+0x3BDF08, 0x2B3715, 0xA0805C, 0x93805A, 0x921110, 0xD8E80F,
+0xAF806C, 0x4BFFDB, 0x0F9038, 0x761859, 0x15A562, 0xBBCB61,
+0xB989C7, 0xBD4010, 0x04F2D2, 0x277549, 0xF6B6EB, 0xBB22DB,
+0xAA140A, 0x2F2689, 0x768364, 0x333B09, 0x1A940E, 0xAA3A51,
+0xC2A31D, 0xAEEDAF, 0x12265C, 0x4DC26D, 0x9C7A2D, 0x9756C0,
+0x833F03, 0xF6F009, 0x8C402B, 0x99316D, 0x07B439, 0x15200C,
+0x5BC3D8, 0xC492F5, 0x4BADC6, 0xA5CA4E, 0xCD37A7, 0x36A9E6,
+0x9492AB, 0x6842DD, 0xDE6319, 0xEF8C76, 0x528B68, 0x37DBFC,
+0xABA1AE, 0x3115DF, 0xA1AE00, 0xDAFB0C, 0x664D64, 0xB705ED,
+0x306529, 0xBF5657, 0x3AFF47, 0xB9F96A, 0xF3BE75, 0xDF9328,
+0x3080AB, 0xF68C66, 0x15CB04, 0x0622FA, 0x1DE4D9, 0xA4B33D,
+0x8F1B57, 0x09CD36, 0xE9424E, 0xA4BE13, 0xB52333, 0x1AAAF0,
+0xA8654F, 0xA5C1D2, 0x0F3F0B, 0xCD785B, 0x76F923, 0x048B7B,
+0x721789, 0x53A6C6, 0xE26E6F, 0x00EBEF, 0x584A9B, 0xB7DAC4,
+0xBA66AA, 0xCFCF76, 0x1D02D1, 0x2DF1B1, 0xC1998C, 0x77ADC3,
+0xDA4886, 0xA05DF7, 0xF480C6, 0x2FF0AC, 0x9AECDD, 0xBC5C3F,
+0x6DDED0, 0x1FC790, 0xB6DB2A, 0x3A25A3, 0x9AAF00, 0x9353AD,
+0x0457B6, 0xB42D29, 0x7E804B, 0xA707DA, 0x0EAA76, 0xA1597B,
+0x2A1216, 0x2DB7DC, 0xFDE5FA, 0xFEDB89, 0xFDBE89, 0x6C76E4,
+0xFCA906, 0x70803E, 0x156E85, 0xFF87FD, 0x073E28, 0x336761,
+0x86182A, 0xEABD4D, 0xAFE7B3, 0x6E6D8F, 0x396795, 0x5BBF31,
+0x48D784, 0x16DF30, 0x432DC7, 0x356125, 0xCE70C9, 0xB8CB30,
+0xFD6CBF, 0xA200A4, 0xE46C05, 0xA0DD5A, 0x476F21, 0xD21262,
+0x845CB9, 0x496170, 0xE0566B, 0x015299, 0x375550, 0xB7D51E,
+0xC4F133, 0x5F6E13, 0xE4305D, 0xA92E85, 0xC3B21D, 0x3632A1,
+0xA4B708, 0xD4B1EA, 0x21F716, 0xE4698F, 0x77FF27, 0x80030C,
+0x2D408D, 0xA0CD4F, 0x99A520, 0xD3A2B3, 0x0A5D2F, 0x42F9B4,
+0xCBDA11, 0xD0BE7D, 0xC1DB9B, 0xBD17AB, 0x81A2CA, 0x5C6A08,
+0x17552E, 0x550027, 0xF0147F, 0x8607E1, 0x640B14, 0x8D4196,
+0xDEBE87, 0x2AFDDA, 0xB6256B, 0x34897B, 0xFEF305, 0x9EBFB9,
+0x4F6A68, 0xA82A4A, 0x5AC44F, 0xBCF82D, 0x985AD7, 0x95C7F4,
+0x8D4D0D, 0xA63A20, 0x5F57A4, 0xB13F14, 0x953880, 0x0120CC,
+0x86DD71, 0xB6DEC9, 0xF560BF, 0x11654D, 0x6B0701, 0xACB08C,
+0xD0C0B2, 0x485551, 0x0EFB1E, 0xC37295, 0x3B06A3, 0x3540C0,
+0x7BDC06, 0xCC45E0, 0xFA294E, 0xC8CAD6, 0x41F3E8, 0xDE647C,
+0xD8649B, 0x31BED9, 0xC397A4, 0xD45877, 0xC5E369, 0x13DAF0,
+0x3C3ABA, 0x461846, 0x5F7555, 0xF5BDD2, 0xC6926E, 0x5D2EAC,
+0xED440E, 0x423E1C, 0x87C461, 0xE9FD29, 0xF3D6E7, 0xCA7C22,
+0x35916F, 0xC5E008, 0x8DD7FF, 0xE26A6E, 0xC6FDB0, 0xC10893,
+0x745D7C, 0xB2AD6B, 0x9D6ECD, 0x7B723E, 0x6A11C6, 0xA9CFF7,
+0xDF7329, 0xBAC9B5, 0x5100B7, 0x0DB2E2, 0x24BA74, 0x607DE5,
+0x8AD874, 0x2C150D, 0x0C1881, 0x94667E, 0x162901, 0x767A9F,
+0xBEFDFD, 0xEF4556, 0x367ED9, 0x13D9EC, 0xB9BA8B, 0xFC97C4,
+0x27A831, 0xC36EF1, 0x36C594, 0x56A8D8, 0xB5A8B4, 0x0ECCCF,
+0x2D8912, 0x34576F, 0x89562C, 0xE3CE99, 0xB920D6, 0xAA5E6B,
+0x9C2A3E, 0xCC5F11, 0x4A0BFD, 0xFBF4E1, 0x6D3B8E, 0x2C86E2,
+0x84D4E9, 0xA9B4FC, 0xD1EEEF, 0xC9352E, 0x61392F, 0x442138,
+0xC8D91B, 0x0AFC81, 0x6A4AFB, 0xD81C2F, 0x84B453, 0x8C994E,
+0xCC2254, 0xDC552A, 0xD6C6C0, 0x96190B, 0xB8701A, 0x649569,
+0x605A26, 0xEE523F, 0x0F117F, 0x11B5F4, 0xF5CBFC, 0x2DBC34,
+0xEEBC34, 0xCC5DE8, 0x605EDD, 0x9B8E67, 0xEF3392, 0xB817C9,
+0x9B5861, 0xBC57E1, 0xC68351, 0x103ED8, 0x4871DD, 0xDD1C2D,
+0xA118AF, 0x462C21, 0xD7F359, 0x987AD9, 0xC0549E, 0xFA864F,
+0xFC0656, 0xAE79E5, 0x362289, 0x22AD38, 0xDC9367, 0xAAE855,
+0x382682, 0x9BE7CA, 0xA40D51, 0xB13399, 0x0ED7A9, 0x480569,
+0xF0B265, 0xA7887F, 0x974C88, 0x36D1F9, 0xB39221, 0x4A827B,
+0x21CF98, 0xDC9F40, 0x5547DC, 0x3A74E1, 0x42EB67, 0xDF9DFE,
+0x5FD45E, 0xA4677B, 0x7AACBA, 0xA2F655, 0x23882B, 0x55BA41,
+0x086E59, 0x862A21, 0x834739, 0xE6E389, 0xD49EE5, 0x40FB49,
+0xE956FF, 0xCA0F1C, 0x8A59C5, 0x2BFA94, 0xC5C1D3, 0xCFC50F,
+0xAE5ADB, 0x86C547, 0x624385, 0x3B8621, 0x94792C, 0x876110,
+0x7B4C2A, 0x1A2C80, 0x12BF43, 0x902688, 0x893C78, 0xE4C4A8,
+0x7BDBE5, 0xC23AC4, 0xEAF426, 0x8A67F7, 0xBF920D, 0x2BA365,
+0xB1933D, 0x0B7CBD, 0xDC51A4, 0x63DD27, 0xDDE169, 0x19949A,
+0x9529A8, 0x28CE68, 0xB4ED09, 0x209F44, 0xCA984E, 0x638270,
+0x237C7E, 0x32B90F, 0x8EF5A7, 0xE75614, 0x08F121, 0x2A9DB5,
+0x4D7E6F, 0x5119A5, 0xABF9B5, 0xD6DF82, 0x61DD96, 0x023616,
+0x9F3AC4, 0xA1A283, 0x6DED72, 0x7A8D39, 0xA9B882, 0x5C326B,
+0x5B2746, 0xED3400, 0x7700D2, 0x55F4FC, 0x4D5901, 0x8071E0,
+#endif
+};
+
+static const double PIo2[] = {
+  1.57079625129699707031e+00, /* 0x3FF921FB, 0x40000000 */
+  7.54978941586159635335e-08, /* 0x3E74442D, 0x00000000 */
+  5.39030252995776476554e-15, /* 0x3CF84698, 0x80000000 */
+  3.28200341580791294123e-22, /* 0x3B78CC51, 0x60000000 */
+  1.27065575308067607349e-29, /* 0x39F01B83, 0x80000000 */
+  1.22933308981111328932e-36, /* 0x387A2520, 0x40000000 */
+  2.73370053816464559624e-44, /* 0x36E38222, 0x80000000 */
+  2.16741683877804819444e-51, /* 0x3569F31D, 0x00000000 */
+};
+
+int __rem_pio2_large(double *x, double *y, int e0, int nx, int prec)
+{
+	int32_t jz,jx,jv,jp,jk,carry,n,iq[20],i,j,k,m,q0,ih;
+	double z,fw,f[20],fq[20],q[20];
+
+	/* initialize jk*/
+	jk = init_jk[prec];
+	jp = jk;
+
+	/* determine jx,jv,q0, note that 3>q0 */
+	jx = nx-1;
+	jv = (e0-3)/24;  if(jv<0) jv=0;
+	q0 = e0-24*(jv+1);
+
+	/* set up f[0] to f[jx+jk] where f[jx+jk] = ipio2[jv+jk] */
+	j = jv-jx; m = jx+jk;
+	for (i=0; i<=m; i++,j++)
+		f[i] = j<0 ? 0.0 : (double)ipio2[j];
+
+	/* compute q[0],q[1],...q[jk] */
+	for (i=0; i<=jk; i++) {
+		for (j=0,fw=0.0; j<=jx; j++)
+			fw += x[j]*f[jx+i-j];
+		q[i] = fw;
+	}
+
+	jz = jk;
+recompute:
+	/* distill q[] into iq[] reversingly */
+	for (i=0,j=jz,z=q[jz]; j>0; i++,j--) {
+		fw    = (double)(int32_t)(0x1p-24*z);
+		iq[i] = (int32_t)(z - 0x1p24*fw);
+		z     = q[j-1]+fw;
+	}
+
+	/* compute n */
+	z  = scalbn(z,q0);       /* actual value of z */
+	z -= 8.0*floor(z*0.125); /* trim off integer >= 8 */
+	n  = (int32_t)z;
+	z -= (double)n;
+	ih = 0;
+	if (q0 > 0) {  /* need iq[jz-1] to determine n */
+		i  = iq[jz-1]>>(24-q0); n += i;
+		iq[jz-1] -= i<<(24-q0);
+		ih = iq[jz-1]>>(23-q0);
+	}
+	else if (q0 == 0) ih = iq[jz-1]>>23;
+	else if (z >= 0.5) ih = 2;
+
+	if (ih > 0) {  /* q > 0.5 */
+		n += 1; carry = 0;
+		for (i=0; i<jz; i++) {  /* compute 1-q */
+			j = iq[i];
+			if (carry == 0) {
+				if (j != 0) {
+					carry = 1;
+					iq[i] = 0x1000000 - j;
+				}
+			} else
+				iq[i] = 0xffffff - j;
+		}
+		if (q0 > 0) {  /* rare case: chance is 1 in 12 */
+			switch(q0) {
+			case 1:
+				iq[jz-1] &= 0x7fffff; break;
+			case 2:
+				iq[jz-1] &= 0x3fffff; break;
+			}
+		}
+		if (ih == 2) {
+			z = 1.0 - z;
+			if (carry != 0)
+				z -= scalbn(1.0,q0);
+		}
+	}
+
+	/* check if recomputation is needed */
+	if (z == 0.0) {
+		j = 0;
+		for (i=jz-1; i>=jk; i--) j |= iq[i];
+		if (j == 0) {  /* need recomputation */
+			for (k=1; iq[jk-k]==0; k++);  /* k = no. of terms needed */
+
+			for (i=jz+1; i<=jz+k; i++) {  /* add q[jz+1] to q[jz+k] */
+				f[jx+i] = (double)ipio2[jv+i];
+				for (j=0,fw=0.0; j<=jx; j++)
+					fw += x[j]*f[jx+i-j];
+				q[i] = fw;
+			}
+			jz += k;
+			goto recompute;
+		}
+	}
+
+	/* chop off zero terms */
+	if (z == 0.0) {
+		jz -= 1;
+		q0 -= 24;
+		while (iq[jz] == 0) {
+			jz--;
+			q0 -= 24;
+		}
+	} else { /* break z into 24-bit if necessary */
+		z = scalbn(z,-q0);
+		if (z >= 0x1p24) {
+			fw = (double)(int32_t)(0x1p-24*z);
+			iq[jz] = (int32_t)(z - 0x1p24*fw);
+			jz += 1;
+			q0 += 24;
+			iq[jz] = (int32_t)fw;
+		} else
+			iq[jz] = (int32_t)z;
+	}
+
+	/* convert integer "bit" chunk to floating-point value */
+	fw = scalbn(1.0,q0);
+	for (i=jz; i>=0; i--) {
+		q[i] = fw*(double)iq[i];
+		fw *= 0x1p-24;
+	}
+
+	/* compute PIo2[0,...,jp]*q[jz,...,0] */
+	for(i=jz; i>=0; i--) {
+		for (fw=0.0,k=0; k<=jp && k<=jz-i; k++)
+			fw += PIo2[k]*q[i+k];
+		fq[jz-i] = fw;
+	}
+
+	/* compress fq[] into y[] */
+	switch(prec) {
+	case 0:
+		fw = 0.0;
+		for (i=jz; i>=0; i--)
+			fw += fq[i];
+		y[0] = ih==0 ? fw : -fw;
+		break;
+	case 1:
+	case 2:
+		fw = 0.0;
+		for (i=jz; i>=0; i--)
+			fw += fq[i];
+		// TODO: drop excess precision here once double_t is used
+		fw = (double)fw;
+		y[0] = ih==0 ? fw : -fw;
+		fw = fq[0]-fw;
+		for (i=1; i<=jz; i++)
+			fw += fq[i];
+		y[1] = ih==0 ? fw : -fw;
+		break;
+	case 3:  /* painful */
+		for (i=jz; i>0; i--) {
+			fw      = fq[i-1]+fq[i];
+			fq[i]  += fq[i-1]-fw;
+			fq[i-1] = fw;
+		}
+		for (i=jz; i>1; i--) {
+			fw      = fq[i-1]+fq[i];
+			fq[i]  += fq[i-1]-fw;
+			fq[i-1] = fw;
+		}
+		for (fw=0.0,i=jz; i>=2; i--)
+			fw += fq[i];
+		if (ih==0) {
+			y[0] =  fq[0]; y[1] =  fq[1]; y[2] =  fw;
+		} else {
+			y[0] = -fq[0]; y[1] = -fq[1]; y[2] = -fw;
+		}
+	}
+	return n&7;
+}
diff --git a/cstdlib/src/__rem_pio2f.c b/cstdlib/src/__rem_pio2f.c
new file mode 100644
index 0000000..e676564
--- /dev/null
+++ b/cstdlib/src/__rem_pio2f.c
@@ -0,0 +1,86 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/e_rem_pio2f.c */
+/*
+ * Conversion to float by Ian Lance Taylor, Cygnus Support, ian@cygnus.com.
+ * Debugged and optimized by Bruce D. Evans.
+ */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+/* __rem_pio2f(x,y)
+ *
+ * return the remainder of x rem pi/2 in *y
+ * use double precision for everything except passing x
+ * use __rem_pio2_large() for large x
+ */
+
+#include "libm.h"
+
+#if FLT_EVAL_METHOD==0 || FLT_EVAL_METHOD==1
+#define EPS DBL_EPSILON
+#elif FLT_EVAL_METHOD==2
+#define EPS LDBL_EPSILON
+#endif
+
+/*
+ * invpio2:  53 bits of 2/pi
+ * pio2_1:   first 25 bits of pi/2
+ * pio2_1t:  pi/2 - pio2_1
+ */
+static const double
+toint   = 1.5/EPS,
+pio4    = 0x1.921fb6p-1,
+invpio2 = 6.36619772367581382433e-01, /* 0x3FE45F30, 0x6DC9C883 */
+pio2_1  = 1.57079631090164184570e+00, /* 0x3FF921FB, 0x50000000 */
+pio2_1t = 1.58932547735281966916e-08; /* 0x3E5110b4, 0x611A6263 */
+
+int __rem_pio2f(float x, double *y)
+{
+	union {float f; uint32_t i;} u = {x};
+	double tx[1],ty[1];
+	double_t fn;
+	uint32_t ix;
+	int n, sign, e0;
+
+	ix = u.i & 0x7fffffff;
+	/* 25+53 bit pi is good enough for medium size */
+	if (ix < 0x4dc90fdb) {  /* |x| ~< 2^28*(pi/2), medium size */
+		/* Use a specialized rint() to get fn. */
+		fn = (double_t)x*invpio2 + toint - toint;
+		n  = (int32_t)fn;
+		*y = x - fn*pio2_1 - fn*pio2_1t;
+		/* Matters with directed rounding. */
+		if (predict_false(*y < -pio4)) {
+			n--;
+			fn--;
+			*y = x - fn*pio2_1 - fn*pio2_1t;
+		} else if (predict_false(*y > pio4)) {
+			n++;
+			fn++;
+			*y = x - fn*pio2_1 - fn*pio2_1t;
+		}
+		return n;
+	}
+	if(ix>=0x7f800000) {  /* x is inf or NaN */
+		*y = x-x;
+		return 0;
+	}
+	/* scale x into [2^23, 2^24-1] */
+	sign = u.i>>31;
+	e0 = (ix>>23) - (0x7f+23);  /* e0 = ilogb(|x|)-23, positive */
+	u.i = ix - (e0<<23);
+	tx[0] = u.f;
+	n  =  __rem_pio2_large(tx,ty,e0,1,0);
+	if (sign) {
+		*y = -ty[0];
+		return -n;
+	}
+	*y = ty[0];
+	return n;
+}
diff --git a/cstdlib/src/__sin.c b/cstdlib/src/__sin.c
new file mode 100644
index 0000000..4030949
--- /dev/null
+++ b/cstdlib/src/__sin.c
@@ -0,0 +1,64 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/k_sin.c */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunSoft, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+/* __sin( x, y, iy)
+ * kernel sin function on ~[-pi/4, pi/4] (except on -0), pi/4 ~ 0.7854
+ * Input x is assumed to be bounded by ~pi/4 in magnitude.
+ * Input y is the tail of x.
+ * Input iy indicates whether y is 0. (if iy=0, y assume to be 0).
+ *
+ * Algorithm
+ *      1. Since sin(-x) = -sin(x), we need only to consider positive x.
+ *      2. Callers must return sin(-0) = -0 without calling here since our
+ *         odd polynomial is not evaluated in a way that preserves -0.
+ *         Callers may do the optimization sin(x) ~ x for tiny x.
+ *      3. sin(x) is approximated by a polynomial of degree 13 on
+ *         [0,pi/4]
+ *                               3            13
+ *              sin(x) ~ x + S1*x + ... + S6*x
+ *         where
+ *
+ *      |sin(x)         2     4     6     8     10     12  |     -58
+ *      |----- - (1+S1*x +S2*x +S3*x +S4*x +S5*x  +S6*x   )| <= 2
+ *      |  x                                               |
+ *
+ *      4. sin(x+y) = sin(x) + sin'(x')*y
+ *                  ~ sin(x) + (1-x*x/2)*y
+ *         For better accuracy, let
+ *                   3      2      2      2      2
+ *              r = x *(S2+x *(S3+x *(S4+x *(S5+x *S6))))
+ *         then                   3    2
+ *              sin(x) = x + (S1*x + (x *(r-y/2)+y))
+ */
+
+#include "libm.h"
+
+static const double
+S1  = -1.66666666666666324348e-01, /* 0xBFC55555, 0x55555549 */
+S2  =  8.33333333332248946124e-03, /* 0x3F811111, 0x1110F8A6 */
+S3  = -1.98412698298579493134e-04, /* 0xBF2A01A0, 0x19C161D5 */
+S4  =  2.75573137070700676789e-06, /* 0x3EC71DE3, 0x57B1FE7D */
+S5  = -2.50507602534068634195e-08, /* 0xBE5AE5E6, 0x8A2B9CEB */
+S6  =  1.58969099521155010221e-10; /* 0x3DE5D93A, 0x5ACFD57C */
+
+double __sin(double x, double y, int iy)
+{
+	double_t z,r,v,w;
+
+	z = x*x;
+	w = z*z;
+	r = S2 + z*(S3 + z*S4) + z*w*(S5 + z*S6);
+	v = z*x;
+	if (iy == 0)
+		return x + v*(S1 + z*r);
+	else
+		return x - ((z*(0.5*y - v*r) - y) - v*S1);
+}
diff --git a/cstdlib/src/__sindf.c b/cstdlib/src/__sindf.c
new file mode 100644
index 0000000..8fec2a3
--- /dev/null
+++ b/cstdlib/src/__sindf.c
@@ -0,0 +1,36 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/k_sinf.c */
+/*
+ * Conversion to float by Ian Lance Taylor, Cygnus Support, ian@cygnus.com.
+ * Optimized by Bruce D. Evans.
+ */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+
+#include "libm.h"
+
+/* |sin(x)/x - s(x)| < 2**-37.5 (~[-4.89e-12, 4.824e-12]). */
+static const double
+S1 = -0x15555554cbac77.0p-55, /* -0.166666666416265235595 */
+S2 =  0x111110896efbb2.0p-59, /*  0.0083333293858894631756 */
+S3 = -0x1a00f9e2cae774.0p-65, /* -0.000198393348360966317347 */
+S4 =  0x16cd878c3b46a7.0p-71; /*  0.0000027183114939898219064 */
+
+float __sindf(double x)
+{
+	double_t r, s, w, z;
+
+	/* Try to optimize for parallel evaluation as in __tandf.c. */
+	z = x*x;
+	w = z*z;
+	r = S3 + z*S4;
+	s = z*x;
+	return (x + s*(S1 + z*S2)) + s*w*r;
+}
diff --git a/cstdlib/src/cos.c b/cstdlib/src/cos.c
new file mode 100644
index 0000000..ee97f68
--- /dev/null
+++ b/cstdlib/src/cos.c
@@ -0,0 +1,77 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/s_cos.c */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+/* cos(x)
+ * Return cosine function of x.
+ *
+ * kernel function:
+ *      __sin           ... sine function on [-pi/4,pi/4]
+ *      __cos           ... cosine function on [-pi/4,pi/4]
+ *      __rem_pio2      ... argument reduction routine
+ *
+ * Method.
+ *      Let S,C and T denote the sin, cos and tan respectively on
+ *      [-PI/4, +PI/4]. Reduce the argument x to y1+y2 = x-k*pi/2
+ *      in [-pi/4 , +pi/4], and let n = k mod 4.
+ *      We have
+ *
+ *          n        sin(x)      cos(x)        tan(x)
+ *     ----------------------------------------------------------
+ *          0          S           C             T
+ *          1          C          -S            -1/T
+ *          2         -S          -C             T
+ *          3         -C           S            -1/T
+ *     ----------------------------------------------------------
+ *
+ * Special cases:
+ *      Let trig be any of sin, cos, or tan.
+ *      trig(+-INF)  is NaN, with signals;
+ *      trig(NaN)    is that NaN;
+ *
+ * Accuracy:
+ *      TRIG(x) returns trig(x) nearly rounded
+ */
+
+#include "libm.h"
+
+double cos(double x)
+{
+	double y[2];
+	uint32_t ix;
+	unsigned n;
+
+	GET_HIGH_WORD(ix, x);
+	ix &= 0x7fffffff;
+
+	/* |x| ~< pi/4 */
+	if (ix <= 0x3fe921fb) {
+		if (ix < 0x3e46a09e) {  /* |x| < 2**-27 * sqrt(2) */
+			/* raise inexact if x!=0 */
+			FORCE_EVAL(x + 0x1p120f);
+			return 1.0;
+		}
+		return __cos(x, 0);
+	}
+
+	/* cos(Inf or NaN) is NaN */
+	if (ix >= 0x7ff00000)
+		return x-x;
+
+	/* argument reduction */
+	n = __rem_pio2(x, y);
+	switch (n&3) {
+	case 0: return  __cos(y[0], y[1]);
+	case 1: return -__sin(y[0], y[1], 1);
+	case 2: return -__cos(y[0], y[1]);
+	default:
+		return  __sin(y[0], y[1], 1);
+	}
+}
diff --git a/cstdlib/src/cosf.c b/cstdlib/src/cosf.c
new file mode 100644
index 0000000..23f3e5b
--- /dev/null
+++ b/cstdlib/src/cosf.c
@@ -0,0 +1,78 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/s_cosf.c */
+/*
+ * Conversion to float by Ian Lance Taylor, Cygnus Support, ian@cygnus.com.
+ * Optimized by Bruce D. Evans.
+ */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+
+#include "libm.h"
+
+/* Small multiples of pi/2 rounded to double precision. */
+static const double
+c1pio2 = 1*M_PI_2, /* 0x3FF921FB, 0x54442D18 */
+c2pio2 = 2*M_PI_2, /* 0x400921FB, 0x54442D18 */
+c3pio2 = 3*M_PI_2, /* 0x4012D97C, 0x7F3321D2 */
+c4pio2 = 4*M_PI_2; /* 0x401921FB, 0x54442D18 */
+
+float cosf(float x)
+{
+	double y;
+	uint32_t ix;
+	unsigned n, sign;
+
+	GET_FLOAT_WORD(ix, x);
+	sign = ix >> 31;
+	ix &= 0x7fffffff;
+
+	if (ix <= 0x3f490fda) {  /* |x| ~<= pi/4 */
+		if (ix < 0x39800000) {  /* |x| < 2**-12 */
+			/* raise inexact if x != 0 */
+			FORCE_EVAL(x + 0x1p120f);
+			return 1.0f;
+		}
+		return __cosdf(x);
+	}
+	if (ix <= 0x407b53d1) {  /* |x| ~<= 5*pi/4 */
+		if (ix > 0x4016cbe3)  /* |x|  ~> 3*pi/4 */
+			return -__cosdf(sign ? x+c2pio2 : x-c2pio2);
+		else {
+			if (sign)
+				return __sindf(x + c1pio2);
+			else
+				return __sindf(c1pio2 - x);
+		}
+	}
+	if (ix <= 0x40e231d5) {  /* |x| ~<= 9*pi/4 */
+		if (ix > 0x40afeddf)  /* |x| ~> 7*pi/4 */
+			return __cosdf(sign ? x+c4pio2 : x-c4pio2);
+		else {
+			if (sign)
+				return __sindf(-x - c3pio2);
+			else
+				return __sindf(x - c3pio2);
+		}
+	}
+
+	/* cos(Inf or NaN) is NaN */
+	if (ix >= 0x7f800000)
+		return x-x;
+
+	/* general argument reduction needed */
+	n = __rem_pio2f(x,&y);
+	switch (n&3) {
+	case 0: return  __cosdf(y);
+	case 1: return  __sindf(-y);
+	case 2: return -__cosdf(y);
+	default:
+		return  __sindf(y);
+	}
+}
diff --git a/cstdlib/src/libm.h b/cstdlib/src/libm.h
index f54ca1b..a310f59 100644
--- a/cstdlib/src/libm.h
+++ b/cstdlib/src/libm.h
@@ -108,6 +108,27 @@ do {                                              \
 #define SET_LOW_WORD(d,lo)                        \
   INSERT_WORDS(d, asuint64(d)>>32, lo)
 
+#define GET_FLOAT_WORD(w,d)                       \
+do {                                              \
+  (w) = asuint(d);                                \
+} while (0)
+
+#define SET_FLOAT_WORD(d,w)                       \
+do {                                              \
+  (d) = asfloat(w);                               \
+} while (0)
+
+int    __rem_pio2_large(double*,double*,int,int,int);
+
+int    __rem_pio2(double,double*);
+double __sin(double,double,int);
+double __cos(double,double);
+
+int    __rem_pio2f(float,double*);
+float  __sindf(double);
+float  __cosdf(double);
+
+float __math_invalidf(float);
 double __math_xflow(uint32_t, double);
 double __math_uflow(uint32_t);
 double __math_oflow(uint32_t);
diff --git a/cstdlib/src/scalbn.c b/cstdlib/src/scalbn.c
new file mode 100644
index 0000000..182f561
--- /dev/null
+++ b/cstdlib/src/scalbn.c
@@ -0,0 +1,33 @@
+#include <math.h>
+#include <stdint.h>
+
+double scalbn(double x, int n)
+{
+	union {double f; uint64_t i;} u;
+	double_t y = x;
+
+	if (n > 1023) {
+		y *= 0x1p1023;
+		n -= 1023;
+		if (n > 1023) {
+			y *= 0x1p1023;
+			n -= 1023;
+			if (n > 1023)
+				n = 1023;
+		}
+	} else if (n < -1022) {
+		/* make sure final n < -53 to avoid double
+		   rounding in the subnormal range */
+		y *= 0x1p-1022 * 0x1p53;
+		n += 1022 - 53;
+		if (n < -1022) {
+			y *= 0x1p-1022 * 0x1p53;
+			n += 1022 - 53;
+			if (n < -1022)
+				n = -1022;
+		}
+	}
+	u.i = (uint64_t)(0x3ff+n)<<52;
+	x = y * u.f;
+	return x;
+}
diff --git a/cstdlib/src/sin.c b/cstdlib/src/sin.c
new file mode 100644
index 0000000..055e215
--- /dev/null
+++ b/cstdlib/src/sin.c
@@ -0,0 +1,78 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/s_sin.c */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+/* sin(x)
+ * Return sine function of x.
+ *
+ * kernel function:
+ *      __sin            ... sine function on [-pi/4,pi/4]
+ *      __cos            ... cose function on [-pi/4,pi/4]
+ *      __rem_pio2       ... argument reduction routine
+ *
+ * Method.
+ *      Let S,C and T denote the sin, cos and tan respectively on
+ *      [-PI/4, +PI/4]. Reduce the argument x to y1+y2 = x-k*pi/2
+ *      in [-pi/4 , +pi/4], and let n = k mod 4.
+ *      We have
+ *
+ *          n        sin(x)      cos(x)        tan(x)
+ *     ----------------------------------------------------------
+ *          0          S           C             T
+ *          1          C          -S            -1/T
+ *          2         -S          -C             T
+ *          3         -C           S            -1/T
+ *     ----------------------------------------------------------
+ *
+ * Special cases:
+ *      Let trig be any of sin, cos, or tan.
+ *      trig(+-INF)  is NaN, with signals;
+ *      trig(NaN)    is that NaN;
+ *
+ * Accuracy:
+ *      TRIG(x) returns trig(x) nearly rounded
+ */
+
+#include "libm.h"
+
+double sin(double x)
+{
+	double y[2];
+	uint32_t ix;
+	unsigned n;
+
+	/* High word of x. */
+	GET_HIGH_WORD(ix, x);
+	ix &= 0x7fffffff;
+
+	/* |x| ~< pi/4 */
+	if (ix <= 0x3fe921fb) {
+		if (ix < 0x3e500000) {  /* |x| < 2**-26 */
+			/* raise inexact if x != 0 and underflow if subnormal*/
+			FORCE_EVAL(ix < 0x00100000 ? x/0x1p120f : x+0x1p120f);
+			return x;
+		}
+		return __sin(x, 0.0, 0);
+	}
+
+	/* sin(Inf or NaN) is NaN */
+	if (ix >= 0x7ff00000)
+		return x - x;
+
+	/* argument reduction needed */
+	n = __rem_pio2(x, y);
+	switch (n&3) {
+	case 0: return  __sin(y[0], y[1], 1);
+	case 1: return  __cos(y[0], y[1]);
+	case 2: return -__sin(y[0], y[1], 1);
+	default:
+		return -__cos(y[0], y[1]);
+	}
+}
diff --git a/cstdlib/src/sinf.c b/cstdlib/src/sinf.c
new file mode 100644
index 0000000..64e39f5
--- /dev/null
+++ b/cstdlib/src/sinf.c
@@ -0,0 +1,76 @@
+/* origin: FreeBSD /usr/src/lib/msun/src/s_sinf.c */
+/*
+ * Conversion to float by Ian Lance Taylor, Cygnus Support, ian@cygnus.com.
+ * Optimized by Bruce D. Evans.
+ */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+
+#include "libm.h"
+
+/* Small multiples of pi/2 rounded to double precision. */
+static const double
+s1pio2 = 1*M_PI_2, /* 0x3FF921FB, 0x54442D18 */
+s2pio2 = 2*M_PI_2, /* 0x400921FB, 0x54442D18 */
+s3pio2 = 3*M_PI_2, /* 0x4012D97C, 0x7F3321D2 */
+s4pio2 = 4*M_PI_2; /* 0x401921FB, 0x54442D18 */
+
+float sinf(float x)
+{
+	double y;
+	uint32_t ix;
+	int n, sign;
+
+	GET_FLOAT_WORD(ix, x);
+	sign = ix >> 31;
+	ix &= 0x7fffffff;
+
+	if (ix <= 0x3f490fda) {  /* |x| ~<= pi/4 */
+		if (ix < 0x39800000) {  /* |x| < 2**-12 */
+			/* raise inexact if x!=0 and underflow if subnormal */
+			FORCE_EVAL(ix < 0x00800000 ? x/0x1p120f : x+0x1p120f);
+			return x;
+		}
+		return __sindf(x);
+	}
+	if (ix <= 0x407b53d1) {  /* |x| ~<= 5*pi/4 */
+		if (ix <= 0x4016cbe3) {  /* |x| ~<= 3pi/4 */
+			if (sign)
+				return -__cosdf(x + s1pio2);
+			else
+				return __cosdf(x - s1pio2);
+		}
+		return __sindf(sign ? -(x + s2pio2) : -(x - s2pio2));
+	}
+	if (ix <= 0x40e231d5) {  /* |x| ~<= 9*pi/4 */
+		if (ix <= 0x40afeddf) {  /* |x| ~<= 7*pi/4 */
+			if (sign)
+				return __cosdf(x + s3pio2);
+			else
+				return -__cosdf(x - s3pio2);
+		}
+		return __sindf(sign ? x + s4pio2 : x - s4pio2);
+	}
+
+	/* sin(Inf or NaN) is NaN */
+	if (ix >= 0x7f800000)
+		return x - x;
+
+	/* general argument reduction needed */
+	n = __rem_pio2f(x, &y);
+	switch (n&3) {
+	case 0: return  __sindf(y);
+	case 1: return  __cosdf(y);
+	case 2: return  __sindf(-y);
+	default:
+		return -__cosdf(y);
+	}
+}
diff --git a/cstdlib/src/sqrtf.c b/cstdlib/src/sqrtf.c
new file mode 100644
index 0000000..740d81c
--- /dev/null
+++ b/cstdlib/src/sqrtf.c
@@ -0,0 +1,83 @@
+#include <stdint.h>
+#include <math.h>
+#include "libm.h"
+#include "sqrt_data.h"
+
+#define FENV_SUPPORT 1
+
+static inline uint32_t mul32(uint32_t a, uint32_t b)
+{
+	return (uint64_t)a*b >> 32;
+}
+
+/* see sqrt.c for more detailed comments.  */
+
+float sqrtf(float x)
+{
+	uint32_t ix, m, m1, m0, even, ey;
+
+	ix = asuint(x);
+	if (predict_false(ix - 0x00800000 >= 0x7f800000 - 0x00800000)) {
+		/* x < 0x1p-126 or inf or nan.  */
+		if (ix * 2 == 0)
+			return x;
+		if (ix == 0x7f800000)
+			return x;
+		if (ix > 0x7f800000)
+			return __math_invalidf(x);
+		/* x is subnormal, normalize it.  */
+		ix = asuint(x * 0x1p23f);
+		ix -= 23 << 23;
+	}
+
+	/* x = 4^e m; with int e and m in [1, 4).  */
+	even = ix & 0x00800000;
+	m1 = (ix << 8) | 0x80000000;
+	m0 = (ix << 7) & 0x7fffffff;
+	m = even ? m0 : m1;
+
+	/* 2^e is the exponent part of the return value.  */
+	ey = ix >> 1;
+	ey += 0x3f800000 >> 1;
+	ey &= 0x7f800000;
+
+	/* compute r ~ 1/sqrt(m), s ~ sqrt(m) with 2 goldschmidt iterations.  */
+	static const uint32_t three = 0xc0000000;
+	uint32_t r, s, d, u, i;
+	i = (ix >> 17) % 128;
+	r = (uint32_t)__rsqrt_tab[i] << 16;
+	/* |r*sqrt(m) - 1| < 0x1p-8 */
+	s = mul32(m, r);
+	/* |s/sqrt(m) - 1| < 0x1p-8 */
+	d = mul32(s, r);
+	u = three - d;
+	r = mul32(r, u) << 1;
+	/* |r*sqrt(m) - 1| < 0x1.7bp-16 */
+	s = mul32(s, u) << 1;
+	/* |s/sqrt(m) - 1| < 0x1.7bp-16 */
+	d = mul32(s, r);
+	u = three - d;
+	s = mul32(s, u);
+	/* -0x1.03p-28 < s/sqrt(m) - 1 < 0x1.fp-31 */
+	s = (s - 1)>>6;
+	/* s < sqrt(m) < s + 0x1.08p-23 */
+
+	/* compute nearest rounded result.  */
+	uint32_t d0, d1, d2;
+	float y, t;
+	d0 = (m << 16) - s*s;
+	d1 = s - d0;
+	d2 = d1 + s + 1;
+	s += d1 >> 31;
+	s &= 0x007fffff;
+	s |= ey;
+	y = asfloat(s);
+	if (FENV_SUPPORT) {
+		/* handle rounding and inexact exception. */
+		uint32_t tiny = predict_false(d2==0) ? 0 : 0x01000000;
+		tiny |= (d1^d2) & 0x80000000;
+		t = asfloat(tiny);
+		y = eval_as_float(y + t);
+	}
+	return y;
+}
diff --git a/milepost b/milepost
index 36c75a5..59fdc27 160000
--- a/milepost
+++ b/milepost
@@ -1 +1 @@
-Subproject commit 36c75a55dd1d843b29691293410344a9f48a9201
+Subproject commit 59fdc27ac6dd84af4c9ed505213dea0fec641c32
diff --git a/samples/pong/.gitignore b/samples/pong/.gitignore
index ef4fd0d..0742738 100644
--- a/samples/pong/.gitignore
+++ b/samples/pong/.gitignore
@@ -1 +1,3 @@
-Pong
\ No newline at end of file
+Pong
+profile.dtrace
+profile.spall
diff --git a/samples/pong/build.sh b/samples/pong/build.sh
index 1f874cc..cf8ee84 100755
--- a/samples/pong/build.sh
+++ b/samples/pong/build.sh
@@ -11,16 +11,22 @@ else
   CLANG=clang
 fi
 
+STDLIB_DIR=../../cstdlib
+ORCA_SDK_DIR=../../sdk
+MILEPOST_DIR=../../milepost
+
 wasmFlags="--target=wasm32 \
-       --no-standard-libraries \
-       -fno-builtin \
-       -Wl,--no-entry \
-       -Wl,--export-dynamic \
-       -g \
-       -O2 \
-       -mbulk-memory \
-       -D__ORCA__ \
-	   -isystem ../../cstdlib/include -I ../../sdk -I../../milepost/ext -I ../../milepost -I ../../milepost/src -I ../../milepost/src/util -I ../../milepost/src/platform -I../.."
+  --no-standard-libraries \
+  -fno-builtin \
+  -Wl,--no-entry \
+  -Wl,--export-dynamic \
+  -g \
+  -O2 \
+  -mbulk-memory \
+  -D__ORCA__ \
+  -I $STDLIB_DIR/include \
+  -I $ORCA_SDK_DIR \
+  -I $MILEPOST_DIR/ext -I $MILEPOST_DIR -I $MILEPOST_DIR/src -I $MILEPOST_DIR/src/util -I $MILEPOST_DIR/src/platform"
 
 $CLANG $wasmFlags -o ./module.wasm ../../sdk/orca.c ../../cstdlib/src/*.c src/main.c
 
diff --git a/samples/pong/data/ball.png b/samples/pong/data/ball.png
index c918ee5..4294de2 100644
Binary files a/samples/pong/data/ball.png and b/samples/pong/data/ball.png differ
diff --git a/samples/pong/data/underwater.jpg b/samples/pong/data/underwater.jpg
new file mode 100644
index 0000000..8a03630
Binary files /dev/null and b/samples/pong/data/underwater.jpg differ
diff --git a/samples/pong/src/main.c b/samples/pong/src/main.c
index 6f78c3f..85de550 100644
--- a/samples/pong/src/main.c
+++ b/samples/pong/src/main.c
@@ -1,173 +1,172 @@
-/************************************************************//**
-*
-*	@file: wasm_main.cpp
-*	@author: Martin Fouilleul
-*	@date: 14/08/2022
-*	@revision:
-*
-*****************************************************************/
+#include <keys.h>
+#include <graphics.h>
+#include <math.h>
 
-#include"keys.h"
-#include"graphics.h"
+#include <orca.h>
 
-#include"orca.h"
+#define NUM_BLOCKS_PER_ROW 7
+#define NUM_BLOCKS 42 // 7 * 6
 
-#define M_PI 3.14159265358979323846
+#define BLOCKS_WIDTH 810.0f
+#define BLOCK_HEIGHT 30.0f
+#define BLOCKS_PADDING 15.0f
+#define BLOCKS_BOTTOM 300.0f
+const f32 BLOCK_WIDTH = (BLOCKS_WIDTH - ((NUM_BLOCKS_PER_ROW + 1) * BLOCKS_PADDING)) / NUM_BLOCKS_PER_ROW;
 
-extern float cosf(float x);
-extern float sinf(float x);
+#define PADDLE_MAX_LAUNCH_ANGLE 0.7f
 
 const mg_color paddleColor = {1, 0, 0, 1};
-mp_rect paddle = {200, 40, 200, 40};
+mp_rect paddle = {300, 50, 200, 24};
 
 const mg_color ballColor = {1, 1, 0, 1};
-mp_rect ball = {200, 200, 60, 60};
+mp_rect ball = {200, 200, 20, 20};
 
-vec2 velocity = {10, 10};
+vec2 velocity = {5, 5};
+
+// This is upside down from how it will actually be drawn.
+int blockHealth[NUM_BLOCKS] = {
+    0, 1, 1, 1, 1, 1, 0,
+    1, 1, 1, 1, 1, 1, 1,
+    2, 2, 2, 2, 2, 2, 2,
+    2, 2, 2, 2, 2, 2, 2,
+    3, 3, 3, 3, 3, 3, 3,
+    3, 3, 3, 3, 3, 3, 3,
+};
 
 vec2 frameSize = {100, 100};
 
 bool leftDown = false;
 bool rightDown = false;
 
-mg_canvas canvas;
 mg_surface surface;
+mg_canvas canvas;
+mg_image waterImage;
 mg_image ballImage;
 mg_image paddleImage;
 mg_font pongFont;
 
+// TODO(ben): Why is this here? Why isn't it forward-declared by some header?
 mg_surface mg_surface_main(void);
 
+f32 lerp(f32 a, f32 b, f32 t);
+mp_rect blockRect(int i);
+int checkCollision(mp_rect block);
+mg_mat2x3 flipY(mp_rect r);
+mg_mat2x3 flipYAt(vec2 pos);
+
+str8 loadFile(mem_arena* arena, str8 filename) {
+    file_handle file = file_open(filename, FILE_ACCESS_READ, 0);
+    if(file_last_error(file) != IO_OK)
+    {
+        log_error("Couldn't open file %s\n", str8_to_cstring(mem_scratch(), filename));
+    }
+    u64 size = file_size(file);
+    char* buffer = mem_arena_alloc(arena, size);
+    file_read(file, size, buffer);
+    file_close(file);
+    return str8_from_buffer(size, buffer);
+}
+
 ORCA_EXPORT void OnInit(void)
 {
-	//TODO create surface for main window
-	surface = mg_surface_main();
-	canvas = mg_canvas_create();
+    surface = mg_surface_main();
+    canvas = mg_canvas_create();
 
-	log_info("try allocating\n");
+    waterImage = mg_image_create_from_data(surface, loadFile(mem_scratch(), STR8("/underwater.jpg")), false);
+    ballImage = mg_image_create_from_data(surface, loadFile(mem_scratch(), STR8("/ball.png")), false);
+    paddleImage = mg_image_create_from_data(surface, loadFile(mem_scratch(), STR8("/wall.png")), false);
 
-	char* foo = malloc(1024);
-	free(foo);
+    if(mg_image_is_nil(waterImage))
+    {
+		log_error("coulnd't load ball image\n");
+    }
+    if(mg_image_is_nil(ballImage))
+    {
+		log_error("coulnd't load ball image\n");
+    }
+    if(mg_image_is_nil(paddleImage))
+    {
+		log_error("coulnd't load paddle image\n");
+    }
 
-	log_info("allocated and freed 1024 bytes\n");
+    str8 fontStr = loadFile(mem_scratch(), STR8("/Literata-SemiBoldItalic.ttf"));
+    unicode_range ranges[5] = {UNICODE_RANGE_BASIC_LATIN,
+                               UNICODE_RANGE_C1_CONTROLS_AND_LATIN_1_SUPPLEMENT,
+                               UNICODE_RANGE_LATIN_EXTENDED_A,
+                               UNICODE_RANGE_LATIN_EXTENDED_B,
+                               UNICODE_RANGE_SPECIALS};
+    // NOTE(ben): Weird that images are "create from data" but fonts are "create from memory"
+    // TODO: Decide whether we're using strings or explicit pointer + length
+    pongFont = mg_font_create_from_memory(fontStr.len, (byte*)fontStr.ptr, 5, ranges);
 
-	//NOTE: load ball texture
-	{
-		file_handle file = file_open(STR8("/ball.png"), FILE_ACCESS_READ, 0);
-		if(file_last_error(file) != IO_OK)
-		{
-			log_error("Couldn't open file ball.png\n");
-		}
-		u64 size = file_size(file);
-		char* buffer = mem_arena_alloc(mem_scratch(), size);
-		file_read(file, size, buffer);
-		file_close(file);
-		ballImage = mg_image_create_from_data(surface, str8_from_buffer(size, buffer), false);
-	}
-
-	//NOTE: load paddle texture
-	{
-		file_handle file = file_open(STR8("/wall.png"), FILE_ACCESS_READ, 0);
-		if(file_last_error(file) != IO_OK)
-		{
-			log_error("Couldn't open file wall.png\n");
-		}
-		u64 size = file_size(file);
-		char* buffer = mem_arena_alloc(mem_scratch(), size);
-		file_read(file, size, buffer);
-		file_close(file);
-		paddleImage = mg_image_create_from_data(surface, str8_from_buffer(size, buffer), false);
-	}
-
-	//NOTE: load paddle texture
-	{
-		file_handle file = file_open(STR8("/Literata-SemiBoldItalic.ttf"), FILE_ACCESS_READ, 0);
-		if(file_last_error(file) != IO_OK)
-		{
-			log_error("Couldn't open file Literata-SemiBoldItalic.ttf\n");
-		}
-		u64 size = file_size(file);
-		char* buffer = mem_arena_alloc(mem_scratch(), size);
-		file_read(file, size, buffer);
-		file_close(file);
-		unicode_range ranges[5] = {UNICODE_RANGE_BASIC_LATIN,
-	                           UNICODE_RANGE_C1_CONTROLS_AND_LATIN_1_SUPPLEMENT,
-	                           UNICODE_RANGE_LATIN_EXTENDED_A,
-	                           UNICODE_RANGE_LATIN_EXTENDED_B,
-	                           UNICODE_RANGE_SPECIALS};
-		// NOTE(ben): Weird that images are "create from data" but fonts are "create from memory"
-		// TODO: Decide whether we're using strings or explicit pointer + length
-		pongFont = mg_font_create_from_memory(size, (byte*)buffer, 5, ranges);
-	}
-
-	mem_arena_clear(mem_scratch());
+    mem_arena_clear(mem_scratch());
 }
 
 ORCA_EXPORT void OnFrameResize(u32 width, u32 height)
 {
-	log_info("frame resize %u, %u", width, height);
-	frameSize.x = width;
-	frameSize.y = height;
+    log_info("frame resize %u, %u", width, height);
+    frameSize.x = width;
+    frameSize.y = height;
 }
 
 ORCA_EXPORT void OnMouseDown(int button)
 {
-	log_info("mouse down!");
+    log_info("mouse down!");
 }
 
 ORCA_EXPORT void OnKeyDown(int key)
 {
-	if(key == KEY_SPACE)
-	{
-		log_error("(this is just for testing errors)");
-		return;
-	}
-	if(key == KEY_ENTER)
-	{
-		log_warning("(this is just for testing warning)");
-		return;
-	}
+    if(key == KEY_SPACE)
+    {
+        log_error("(this is just for testing errors)");
+        return;
+    }
+    if(key == KEY_ENTER)
+    {
+        log_warning("(this is just for testing warning)");
+        return;
+    }
 
-	log_info("key down: %i", key);
-	if(key == KEY_LEFT)
-	{
-		leftDown = true;
-	}
-	if(key == KEY_RIGHT)
-	{
-		rightDown = true;
-	}
+    log_info("key down: %i", key);
+    if(key == KEY_LEFT)
+    {
+        leftDown = true;
+    }
+    if(key == KEY_RIGHT)
+    {
+        rightDown = true;
+    }
 }
 
 ORCA_EXPORT void OnKeyUp(int key)
 {
-	if(key == KEY_ENTER || key == KEY_SPACE)
-	{
-		return;
-	}
+    if(key == KEY_ENTER || key == KEY_SPACE)
+    {
+        return;
+    }
 
-	log_info("key up: %i", key);
-	if(key == KEY_LEFT)
-	{
-		leftDown = false;
-	}
-	if(key == KEY_RIGHT)
-	{
-		rightDown = false;
-	}
+    log_info("key up: %i", key);
+    if(key == KEY_LEFT)
+    {
+        leftDown = false;
+    }
+    if(key == KEY_RIGHT)
+    {
+        rightDown = false;
+    }
 }
 
 ORCA_EXPORT void OnFrameRefresh(void)
 {
-	f32 aspect = frameSize.x/frameSize.y;
+    f32 aspect = frameSize.x/frameSize.y;
 
     if(leftDown)
     {
-		paddle.x -= 10;
+        paddle.x -= 10;
     }
     else if(rightDown)
     {
-		paddle.x += 10;
+        paddle.x += 10;
     }
     paddle.x = Clamp(paddle.x, 0, frameSize.x - paddle.w);
 
@@ -176,73 +175,291 @@ ORCA_EXPORT void OnFrameRefresh(void)
     ball.x = Clamp(ball.x, 0, frameSize.x - ball.w);
     ball.y = Clamp(ball.y, 0, frameSize.y - ball.h);
 
-    if(ball.x + ball.w >= frameSize.x)
-    {
-		velocity.x = -10;
+    if (ball.x + ball.w >= frameSize.x) {
+        velocity.x = -velocity.x;
     }
-    if(ball.x <= 0)
-    {
-		velocity.x = +10;
+    if (ball.x <= 0) {
+        velocity.x = -velocity.x;
     }
-    if(ball.y + ball.h >= frameSize.y)
-    {
-		velocity.y = -10;
+    if (ball.y + ball.h >= frameSize.y) {
+        velocity.y = -velocity.y;
     }
 
-    if(ball.y <= paddle.y + paddle.h
-       && ball.x+ball.w >= paddle.x
-       && ball.x <= paddle.x + paddle.w
-       && velocity.y < 0)
-    {
-		velocity.y *= -1;
-		ball.y = paddle.y + paddle.h;
+    if (
+        ball.y <= paddle.y + paddle.h
+        && ball.x+ball.w >= paddle.x
+        && ball.x <= paddle.x + paddle.w
+        && velocity.y < 0
+    ) {
+        f32 t = ((ball.x + ball.w/2) - paddle.x) / paddle.w;
+        f32 launchAngle = lerp(-PADDLE_MAX_LAUNCH_ANGLE, PADDLE_MAX_LAUNCH_ANGLE, t);
+        f32 speed = sqrtf(velocity.x*velocity.x + velocity.y*velocity.y);
+        velocity = (vec2){
+            sinf(launchAngle) * speed,
+            cosf(launchAngle) * speed,
+        };
+        ball.y = paddle.y + paddle.h;
 
-		log_info("PONG!");
+        log_info("PONG!");
     }
 
-    if(ball.y <= 0)
+    if (ball.y <= 0) {
+        ball.x = frameSize.x/2. - ball.w;
+        ball.y = frameSize.y/2. - ball.h;
+    }
+
+    for (int i = 0; i < NUM_BLOCKS; i++) {
+        if (blockHealth[i] <= 0) {
+            continue;
+        }
+
+        mp_rect r = blockRect(i);
+        int result = checkCollision(r);
+        if (result) {
+            log_info("Collision! direction=%d", result);
+            blockHealth[i] -= 1;
+
+            f32 vx = velocity.x;
+            f32 vy = velocity.y;
+
+            switch (result) {
+            case 1:
+            case 5:
+                velocity.y = -vy;
+                break;
+            case 3:
+            case 7:
+                velocity.x = -vx;
+                break;
+            case 2:
+            case 6:
+                velocity.x = -vy;
+                velocity.y = -vx;
+                break;
+            case 4:
+            case 8:
+                velocity.x = vy;
+                velocity.y = vx;
+                break;
+            }
+        }
+
+    }
+
+    mg_canvas_set_current(canvas);
+
+    mg_set_color_rgba(10.0f/255.0f, 31.0f/255.0f, 72.0f/255.0f, 1);
+    mg_clear();
+
+    mg_image_draw(waterImage, (mp_rect){0, 0, frameSize.x, frameSize.y});
+
+    mg_mat2x3 yUp = {
+        1, 0, 0,
+        0, -1, frameSize.y,
+    };
+
+    mg_matrix_push(yUp);
     {
-		ball.x = frameSize.x/2. - ball.w;
-		ball.y = frameSize.y/2. - ball.h;
-	}
+        for (int i = 0; i < NUM_BLOCKS; i++) {
+            if (blockHealth[i] <= 0) {
+                continue;
+            }
 
-	mg_canvas_set_current(canvas);
+            mp_rect r = blockRect(i);
+            mg_set_color_rgba(0, 0, 0, 0.2);
+            mg_rounded_rectangle_fill(r.x, r.y-2, r.w, r.h, 4);
+            mg_set_color_rgba(0.9, 0.9, 0.9, 1);
+            mg_rounded_rectangle_fill(r.x, r.y, r.w, r.h, 4);
 
-	mg_set_color_rgba(0, 1, 1, 1);
-	mg_clear();
+            int fontSize = 18;
+            str8 text = str8_pushf(mem_scratch(),
+                "%d", blockHealth[i]
+            );
+            mp_rect textRect = mg_text_bounding_box(pongFont, fontSize, text);
 
-	mg_mat2x3 transform = {1, 0, 0,
-	                       0, -1, frameSize.y};
+            vec2 textPos = {
+                r.x + r.w/2 - textRect.w/2,
+                r.y + 9, // TODO: mg_text_bounding_box is returning extremely wack results for height.
+            };
 
-	mg_matrix_push(transform);
+            mg_set_color_rgba(0, 0, 0, 1);
+            mg_set_font(pongFont);
+            mg_set_font_size(18);
+            mg_move_to(textPos.x, textPos.y);
+            mg_matrix_push(flipYAt(textPos));
+            {
+                mg_text_outlines(text);
+                mg_fill();
+            }
+            mg_matrix_pop();
+        }
 
-	mg_image_draw(paddleImage, paddle);
-	/*
-	mg_set_color(paddleColor);
-	mg_rectangle_fill(paddle.x, paddle.y, paddle.w, paddle.h);
-	*/
-
-	mg_image_draw(ballImage, ball);
-	/*
-	mg_set_color(ballColor);
-	mg_circle_fill(ball.x+ball.w/2, ball.y + ball.w/2, ball.w/2.);
-	*/
+        mg_set_color_rgba(0.9, 0.9, 0.9, 1);
+        mg_rounded_rectangle_fill(paddle.x, paddle.y, paddle.w, paddle.h, 4);
 
+        mg_matrix_push(flipY(ball));
+        {
+            mg_image_draw(ballImage, ball);
+        }
+        mg_matrix_pop();
+    }
     mg_matrix_pop();
 
-	mg_set_color_rgba(0, 0, 0, 1);
-	mg_set_font(pongFont);
-	mg_set_font_size(14);
-	mg_move_to(10, 20);
-
-	str8 text = str8_pushf(mem_scratch(),
-		"wahoo I'm did a text. ball is at x = %f, y = %f",
-		ball.x, ball.y
-	);
-	mg_text_outlines(text);
-	mg_fill();
-
     mg_surface_prepare(surface);
     mg_render(surface, canvas);
     mg_surface_present(surface);
 }
+
+mp_rect blockRect(int i) {
+    int row = i / NUM_BLOCKS_PER_ROW;
+    int col = i % NUM_BLOCKS_PER_ROW;
+    return (mp_rect){
+        BLOCKS_PADDING + (BLOCKS_PADDING + BLOCK_WIDTH) * col,
+        BLOCKS_BOTTOM + (BLOCKS_PADDING + BLOCK_HEIGHT) * row,
+        BLOCK_WIDTH,
+        BLOCK_HEIGHT
+    };
+}
+
+// Returns a cardinal direction 1-8 for the collision with the block, or zero
+// if no collision. 1 is straight up and directions proceed clockwise.
+int checkCollision(mp_rect block) {
+    // Note that all the logic for this game has the origin in the bottom left.
+
+    f32 ballx2 = ball.x + ball.w;
+    f32 bally2 = ball.y + ball.h;
+    f32 blockx2 = block.x + block.w;
+    f32 blocky2 = block.y + block.h;
+
+    if (
+        ballx2 < block.x
+        || blockx2 < ball.x
+        || bally2 < block.y
+        || blocky2 < ball.y
+    ) {
+        // Ball is fully outside block
+        return 0;
+    }
+
+    // if (
+    //     (block.x <= ball.x && ballx2 <= blockx2)
+    //     && (block.y <= ball.y && bally2 <= blocky2)
+    // ) {
+    //     // Ball is fully inside block; do not consider as a collision
+    //     return 0;
+    // }
+
+    // If moving right, the ball can bounce off its top right corner, right
+    // side, or bottom right corner. Corner bounces occur if the block's bottom
+    // left corner is in the ball's top right quadrant, or if the block's top
+    // left corner is in the ball's bottom left quadrant. Otherwise, an edge
+    // bounce occurs if the block's left edge falls in either of the ball's
+    // right quadrants.
+    //
+    // This logic generalizes to other directions.
+    //
+    // We assume significant tunneling can't happen.
+
+    vec2 ballCenter = (vec2){ball.x + ball.w/2, ball.y + ball.h/2};
+    vec2 blockCenter = (vec2){block.x + block.w/2, block.y + block.h/2};
+
+    // Moving right
+    if (velocity.x > 0) {
+        // Ball's top right corner
+        if (
+            ballCenter.x <= block.x && block.x <= ballx2
+            && ballCenter.y <= block.y && block.y <= bally2
+        ) { return 2; }
+
+        // Ball's bottom right corner
+        if (
+            ballCenter.x <= block.x && block.x <= ballx2
+            && ball.y <= blocky2 && blocky2 <= ballCenter.y
+        ) { return 4; }
+
+        // Ball's right edge
+        if (
+            ballCenter.x <= block.x && block.x <= ballx2
+        ) { return 3; }
+    }
+
+    // Moving up
+    if (velocity.y > 0) {
+        // Ball's top left corner
+        if (
+            ball.x <= blockx2 && blockx2 <= ballCenter.x
+            && ballCenter.y <= block.y && block.y <= bally2
+        ) { return 8; }
+
+        // Ball's top right corner
+        if (
+            ballCenter.x <= block.x && block.x <= ballx2
+            && ballCenter.y <= block.y && block.y <= bally2
+        ) { return 2; }
+
+        // Ball's top edge
+        if (
+            ballCenter.y <= block.y && block.y <= bally2
+        ) { return 1; }
+    }
+
+    // Moving left
+    if (velocity.x < 0) {
+        // Ball's bottom left corner
+        if (
+            ball.x <= blockx2 && blockx2 <= ballCenter.x
+            && ball.y <= blocky2 && blocky2 <= ballCenter.y
+        ) { return 6; }
+
+        // Ball's top left corner
+        if (
+            ball.x <= blockx2 && blockx2 <= ballCenter.x
+            && ballCenter.y <= block.y && block.y <= bally2
+        ) { return 8; }
+
+        // Ball's left edge
+        if (
+            ball.x <= blockx2 && blockx2 <= ballCenter.x
+        ) { return 7; }
+    }
+
+    // Moving down
+    if (velocity.y < 0) {
+        // Ball's bottom right corner
+        if (
+            ballCenter.x <= block.x && block.x <= ballx2
+            && ball.y <= blocky2 && blocky2 <= ballCenter.y
+        ) { return 4; }
+
+        // Ball's bottom left corner
+        if (
+            ball.x <= blockx2 && blockx2 <= ballCenter.x
+            && ball.y <= blocky2 && blocky2 <= ballCenter.y
+        ) { return 6; }
+
+        // Ball's bottom edge
+        if (
+            ball.y <= blocky2 && blocky2 <= ballCenter.y
+        ) { return 5; }
+    }
+
+    return 0;
+}
+
+f32 lerp(f32 a, f32 b, f32 t) {
+    return (1 - t) * a + t * b;
+}
+
+mg_mat2x3 flipY(mp_rect r) {
+    return (mg_mat2x3){
+        1, 0, 0,
+        0, -1, 2 * r.y + r.h,
+    };
+}
+
+mg_mat2x3 flipYAt(vec2 pos) {
+    return (mg_mat2x3){
+        1, 0, 0,
+        0, -1, 2 * pos.y,
+    };
+}
diff --git a/samples/pong/trace.sh b/samples/pong/trace.sh
new file mode 100755
index 0000000..48186aa
--- /dev/null
+++ b/samples/pong/trace.sh
@@ -0,0 +1,16 @@
+#!/bin/bash
+
+# You will need dtrace2spall installed:
+#
+#   go install github.com/bvisness/dtrace2spall@latest
+#
+# You will also need `$(go env GOPATH)/bin` on your PATH.
+
+# Run DTrace and save to profile.dtrace
+sudo dtrace -n 'profile-997 /pid == $target/ { @[timestamp, pid, tid, ustack(100)] = count(); }' \
+    -x ustackframes=100 \
+    -o profile.dtrace \
+    -x aggsortkey -x aggsortkeypos=0 \
+    -c ./Pong.app/Contents/MacOS/orca
+# Convert to Spall and save to profile.spall
+cat profile.dtrace | dtrace2spall --freq 997 -o profile.spall --fields=_,pid,tid