
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0 (* (* (- 1.0 ux) maxCos) ux))
(t_1 (sqrt (- 1.0 (* t_0 t_0))))
(t_2 (* (* uy 2.0) PI)))
(+ (+ (* (* (cos t_2) t_1) xi) (* (* (sin t_2) t_1) yi)) (* t_0 zi))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = ((1.0f - ux) * maxCos) * ux;
float t_1 = sqrtf((1.0f - (t_0 * t_0)));
float t_2 = (uy * 2.0f) * ((float) M_PI);
return (((cosf(t_2) * t_1) * xi) + ((sinf(t_2) * t_1) * yi)) + (t_0 * zi);
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(Float32(Float32(Float32(1.0) - ux) * maxCos) * ux) t_1 = sqrt(Float32(Float32(1.0) - Float32(t_0 * t_0))) t_2 = Float32(Float32(uy * Float32(2.0)) * Float32(pi)) return Float32(Float32(Float32(Float32(cos(t_2) * t_1) * xi) + Float32(Float32(sin(t_2) * t_1) * yi)) + Float32(t_0 * zi)) end
function tmp = code(xi, yi, zi, ux, uy, maxCos) t_0 = ((single(1.0) - ux) * maxCos) * ux; t_1 = sqrt((single(1.0) - (t_0 * t_0))); t_2 = (uy * single(2.0)) * single(pi); tmp = (((cos(t_2) * t_1) * xi) + ((sin(t_2) * t_1) * yi)) + (t_0 * zi); end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := \left(\left(1 - ux\right) \cdot maxCos\right) \cdot ux\\
t_1 := \sqrt{1 - t\_0 \cdot t\_0}\\
t_2 := \left(uy \cdot 2\right) \cdot \pi\\
\left(\left(\cos t\_2 \cdot t\_1\right) \cdot xi + \left(\sin t\_2 \cdot t\_1\right) \cdot yi\right) + t\_0 \cdot zi
\end{array}
\end{array}
Herbie found 13 alternatives:
| Alternative | Accuracy | Speedup |
|---|
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0 (* (* (- 1.0 ux) maxCos) ux))
(t_1 (sqrt (- 1.0 (* t_0 t_0))))
(t_2 (* (* uy 2.0) PI)))
(+ (+ (* (* (cos t_2) t_1) xi) (* (* (sin t_2) t_1) yi)) (* t_0 zi))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = ((1.0f - ux) * maxCos) * ux;
float t_1 = sqrtf((1.0f - (t_0 * t_0)));
float t_2 = (uy * 2.0f) * ((float) M_PI);
return (((cosf(t_2) * t_1) * xi) + ((sinf(t_2) * t_1) * yi)) + (t_0 * zi);
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(Float32(Float32(Float32(1.0) - ux) * maxCos) * ux) t_1 = sqrt(Float32(Float32(1.0) - Float32(t_0 * t_0))) t_2 = Float32(Float32(uy * Float32(2.0)) * Float32(pi)) return Float32(Float32(Float32(Float32(cos(t_2) * t_1) * xi) + Float32(Float32(sin(t_2) * t_1) * yi)) + Float32(t_0 * zi)) end
function tmp = code(xi, yi, zi, ux, uy, maxCos) t_0 = ((single(1.0) - ux) * maxCos) * ux; t_1 = sqrt((single(1.0) - (t_0 * t_0))); t_2 = (uy * single(2.0)) * single(pi); tmp = (((cos(t_2) * t_1) * xi) + ((sin(t_2) * t_1) * yi)) + (t_0 * zi); end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := \left(\left(1 - ux\right) \cdot maxCos\right) \cdot ux\\
t_1 := \sqrt{1 - t\_0 \cdot t\_0}\\
t_2 := \left(uy \cdot 2\right) \cdot \pi\\
\left(\left(\cos t\_2 \cdot t\_1\right) \cdot xi + \left(\sin t\_2 \cdot t\_1\right) \cdot yi\right) + t\_0 \cdot zi
\end{array}
\end{array}
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0 (* PI (* 2.0 uy))) (t_1 (* maxCos (- 1.0 ux))))
(fma
(*
(sqrt (fma (- ux 1.0) (* (* (* maxCos ux) maxCos) (* (- 1.0 ux) ux)) 1.0))
(cos t_0))
xi
(fma (* zi t_1) ux (* yi (* (sin t_0) (sin (acos (* t_1 ux)))))))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = ((float) M_PI) * (2.0f * uy);
float t_1 = maxCos * (1.0f - ux);
return fmaf((sqrtf(fmaf((ux - 1.0f), (((maxCos * ux) * maxCos) * ((1.0f - ux) * ux)), 1.0f)) * cosf(t_0)), xi, fmaf((zi * t_1), ux, (yi * (sinf(t_0) * sinf(acosf((t_1 * ux)))))));
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(Float32(pi) * Float32(Float32(2.0) * uy)) t_1 = Float32(maxCos * Float32(Float32(1.0) - ux)) return fma(Float32(sqrt(fma(Float32(ux - Float32(1.0)), Float32(Float32(Float32(maxCos * ux) * maxCos) * Float32(Float32(Float32(1.0) - ux) * ux)), Float32(1.0))) * cos(t_0)), xi, fma(Float32(zi * t_1), ux, Float32(yi * Float32(sin(t_0) * sin(acos(Float32(t_1 * ux))))))) end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := \pi \cdot \left(2 \cdot uy\right)\\
t_1 := maxCos \cdot \left(1 - ux\right)\\
\mathsf{fma}\left(\sqrt{\mathsf{fma}\left(ux - 1, \left(\left(maxCos \cdot ux\right) \cdot maxCos\right) \cdot \left(\left(1 - ux\right) \cdot ux\right), 1\right)} \cdot \cos t\_0, xi, \mathsf{fma}\left(zi \cdot t\_1, ux, yi \cdot \left(\sin t\_0 \cdot \sin \cos^{-1} \left(t\_1 \cdot ux\right)\right)\right)\right)
\end{array}
\end{array}
Initial program 98.9%
lift-+.f32N/A
lift-+.f32N/A
associate-+l+N/A
Applied rewrites99.0%
lift-sin.f32N/A
lift-acos.f32N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
sin-acos-revN/A
lift-*.f32N/A
lift--.f32N/A
lift-sqrt.f3299.0
lift-*.f32N/A
pow2N/A
lower-pow.f3299.0
Applied rewrites99.0%
Applied rewrites99.0%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0 (* (* (- 1.0 ux) maxCos) ux)))
(+
(+
(*
(*
(cos (* (+ uy uy) PI))
(sqrt (fma (* (- ux 1.0) (* ux maxCos)) (* maxCos ux) 1.0)))
xi)
(* (* (sin (* (* uy 2.0) PI)) (sqrt (- 1.0 (* t_0 t_0)))) yi))
(* (* zi (* maxCos (- 1.0 ux))) ux))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = ((1.0f - ux) * maxCos) * ux;
return (((cosf(((uy + uy) * ((float) M_PI))) * sqrtf(fmaf(((ux - 1.0f) * (ux * maxCos)), (maxCos * ux), 1.0f))) * xi) + ((sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf((1.0f - (t_0 * t_0)))) * yi)) + ((zi * (maxCos * (1.0f - ux))) * ux);
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(Float32(Float32(Float32(1.0) - ux) * maxCos) * ux) return Float32(Float32(Float32(Float32(cos(Float32(Float32(uy + uy) * Float32(pi))) * sqrt(fma(Float32(Float32(ux - Float32(1.0)) * Float32(ux * maxCos)), Float32(maxCos * ux), Float32(1.0)))) * xi) + Float32(Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(1.0) - Float32(t_0 * t_0)))) * yi)) + Float32(Float32(zi * Float32(maxCos * Float32(Float32(1.0) - ux))) * ux)) end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := \left(\left(1 - ux\right) \cdot maxCos\right) \cdot ux\\
\left(\left(\cos \left(\left(uy + uy\right) \cdot \pi\right) \cdot \sqrt{\mathsf{fma}\left(\left(ux - 1\right) \cdot \left(ux \cdot maxCos\right), maxCos \cdot ux, 1\right)}\right) \cdot xi + \left(\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - t\_0 \cdot t\_0}\right) \cdot yi\right) + \left(zi \cdot \left(maxCos \cdot \left(1 - ux\right)\right)\right) \cdot ux
\end{array}
\end{array}
Initial program 98.9%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3298.9
Applied rewrites98.9%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.9
Applied rewrites98.9%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3298.9
Applied rewrites98.9%
Taylor expanded in ux around 0
Applied rewrites98.8%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0 (* 2.0 (* uy PI))) (t_1 (* (* (- 1.0 ux) maxCos) ux)))
(if (<= uy 0.0003499999875202775)
(+
(+
(*
(*
(cos (* (+ uy uy) PI))
(sqrt (fma (* (- ux 1.0) (* ux maxCos)) t_1 1.0)))
xi)
(* (* t_0 (sqrt (- 1.0 (* t_1 t_1)))) yi))
(* (* zi (* maxCos (- 1.0 ux))) ux))
(fma xi (cos t_0) (* yi (sin t_0))))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = 2.0f * (uy * ((float) M_PI));
float t_1 = ((1.0f - ux) * maxCos) * ux;
float tmp;
if (uy <= 0.0003499999875202775f) {
tmp = (((cosf(((uy + uy) * ((float) M_PI))) * sqrtf(fmaf(((ux - 1.0f) * (ux * maxCos)), t_1, 1.0f))) * xi) + ((t_0 * sqrtf((1.0f - (t_1 * t_1)))) * yi)) + ((zi * (maxCos * (1.0f - ux))) * ux);
} else {
tmp = fmaf(xi, cosf(t_0), (yi * sinf(t_0)));
}
return tmp;
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(Float32(2.0) * Float32(uy * Float32(pi))) t_1 = Float32(Float32(Float32(Float32(1.0) - ux) * maxCos) * ux) tmp = Float32(0.0) if (uy <= Float32(0.0003499999875202775)) tmp = Float32(Float32(Float32(Float32(cos(Float32(Float32(uy + uy) * Float32(pi))) * sqrt(fma(Float32(Float32(ux - Float32(1.0)) * Float32(ux * maxCos)), t_1, Float32(1.0)))) * xi) + Float32(Float32(t_0 * sqrt(Float32(Float32(1.0) - Float32(t_1 * t_1)))) * yi)) + Float32(Float32(zi * Float32(maxCos * Float32(Float32(1.0) - ux))) * ux)); else tmp = fma(xi, cos(t_0), Float32(yi * sin(t_0))); end return tmp end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := 2 \cdot \left(uy \cdot \pi\right)\\
t_1 := \left(\left(1 - ux\right) \cdot maxCos\right) \cdot ux\\
\mathbf{if}\;uy \leq 0.0003499999875202775:\\
\;\;\;\;\left(\left(\cos \left(\left(uy + uy\right) \cdot \pi\right) \cdot \sqrt{\mathsf{fma}\left(\left(ux - 1\right) \cdot \left(ux \cdot maxCos\right), t\_1, 1\right)}\right) \cdot xi + \left(t\_0 \cdot \sqrt{1 - t\_1 \cdot t\_1}\right) \cdot yi\right) + \left(zi \cdot \left(maxCos \cdot \left(1 - ux\right)\right)\right) \cdot ux\\
\mathbf{else}:\\
\;\;\;\;\mathsf{fma}\left(xi, \cos t\_0, yi \cdot \sin t\_0\right)\\
\end{array}
\end{array}
if uy < 3.49999988e-4Initial program 99.3%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3299.3
Applied rewrites99.3%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3299.3
Applied rewrites99.3%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3299.3
Applied rewrites99.3%
Taylor expanded in uy around 0
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f3299.1
Applied rewrites99.1%
if 3.49999988e-4 < uy Initial program 98.2%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3298.2
Applied rewrites98.2%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.2
Applied rewrites98.2%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3298.2
Applied rewrites98.2%
Taylor expanded in ux around 0
lower-fma.f32N/A
lower-cos.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f32N/A
lower-*.f32N/A
lower-sin.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f3290.3
Applied rewrites90.3%
(FPCore (xi yi zi ux uy maxCos) :precision binary32 (let* ((t_0 (* 2.0 (* uy PI)))) (fma maxCos (* ux (* zi (- 1.0 ux))) (fma xi (cos t_0) (* yi (sin t_0))))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = 2.0f * (uy * ((float) M_PI));
return fmaf(maxCos, (ux * (zi * (1.0f - ux))), fmaf(xi, cosf(t_0), (yi * sinf(t_0))));
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(Float32(2.0) * Float32(uy * Float32(pi))) return fma(maxCos, Float32(ux * Float32(zi * Float32(Float32(1.0) - ux))), fma(xi, cos(t_0), Float32(yi * sin(t_0)))) end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := 2 \cdot \left(uy \cdot \pi\right)\\
\mathsf{fma}\left(maxCos, ux \cdot \left(zi \cdot \left(1 - ux\right)\right), \mathsf{fma}\left(xi, \cos t\_0, yi \cdot \sin t\_0\right)\right)
\end{array}
\end{array}
Initial program 98.9%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3298.9
Applied rewrites98.9%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.9
Applied rewrites98.9%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3298.9
Applied rewrites98.9%
Taylor expanded in maxCos around 0
lower-fma.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f32N/A
lower-fma.f32N/A
Applied rewrites98.7%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0 (* (* (- 1.0 ux) maxCos) ux)) (t_1 (* 2.0 (* uy PI))))
(if (<= uy 0.0003499999875202775)
(+
(+
(*
(sqrt
(+ 1.0 (* (* maxCos maxCos) (* (* ux ux) (* (- 1.0 ux) (- ux 1.0))))))
xi)
(* (* (sin (* (* uy 2.0) PI)) (sqrt (- 1.0 (* t_0 t_0)))) yi))
(* (* zi (* maxCos (- 1.0 ux))) ux))
(fma xi (cos t_1) (* yi (sin t_1))))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = ((1.0f - ux) * maxCos) * ux;
float t_1 = 2.0f * (uy * ((float) M_PI));
float tmp;
if (uy <= 0.0003499999875202775f) {
tmp = ((sqrtf((1.0f + ((maxCos * maxCos) * ((ux * ux) * ((1.0f - ux) * (ux - 1.0f)))))) * xi) + ((sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf((1.0f - (t_0 * t_0)))) * yi)) + ((zi * (maxCos * (1.0f - ux))) * ux);
} else {
tmp = fmaf(xi, cosf(t_1), (yi * sinf(t_1)));
}
return tmp;
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(Float32(Float32(Float32(1.0) - ux) * maxCos) * ux) t_1 = Float32(Float32(2.0) * Float32(uy * Float32(pi))) tmp = Float32(0.0) if (uy <= Float32(0.0003499999875202775)) tmp = Float32(Float32(Float32(sqrt(Float32(Float32(1.0) + Float32(Float32(maxCos * maxCos) * Float32(Float32(ux * ux) * Float32(Float32(Float32(1.0) - ux) * Float32(ux - Float32(1.0))))))) * xi) + Float32(Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(1.0) - Float32(t_0 * t_0)))) * yi)) + Float32(Float32(zi * Float32(maxCos * Float32(Float32(1.0) - ux))) * ux)); else tmp = fma(xi, cos(t_1), Float32(yi * sin(t_1))); end return tmp end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := \left(\left(1 - ux\right) \cdot maxCos\right) \cdot ux\\
t_1 := 2 \cdot \left(uy \cdot \pi\right)\\
\mathbf{if}\;uy \leq 0.0003499999875202775:\\
\;\;\;\;\left(\sqrt{1 + \left(maxCos \cdot maxCos\right) \cdot \left(\left(ux \cdot ux\right) \cdot \left(\left(1 - ux\right) \cdot \left(ux - 1\right)\right)\right)} \cdot xi + \left(\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - t\_0 \cdot t\_0}\right) \cdot yi\right) + \left(zi \cdot \left(maxCos \cdot \left(1 - ux\right)\right)\right) \cdot ux\\
\mathbf{else}:\\
\;\;\;\;\mathsf{fma}\left(xi, \cos t\_1, yi \cdot \sin t\_1\right)\\
\end{array}
\end{array}
if uy < 3.49999988e-4Initial program 99.3%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3299.3
Applied rewrites99.3%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3299.3
Applied rewrites99.3%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3299.3
Applied rewrites99.3%
Taylor expanded in uy around 0
lower-sqrt.f32N/A
lower-+.f32N/A
lower-*.f32N/A
unpow2N/A
lower-*.f32N/A
lower-*.f32N/A
unpow2N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f32N/A
lift--.f3298.8
Applied rewrites98.8%
if 3.49999988e-4 < uy Initial program 98.2%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3298.2
Applied rewrites98.2%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.2
Applied rewrites98.2%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3298.2
Applied rewrites98.2%
Taylor expanded in ux around 0
lower-fma.f32N/A
lower-cos.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f32N/A
lower-*.f32N/A
lower-sin.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f3290.3
Applied rewrites90.3%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0 (* 2.0 (* uy PI))))
(if (<= uy 0.0002800000074785203)
(+
(fma
2.0
(*
(* uy (* yi PI))
(sqrt (- 1.0 (* (* maxCos maxCos) (pow (* ux (- 1.0 ux)) 2.0)))))
(*
xi
(sqrt
(+
1.0
(* (* maxCos maxCos) (* (* ux ux) (* (- 1.0 ux) (- ux 1.0))))))))
(* (* zi (* maxCos (- 1.0 ux))) ux))
(fma xi (cos t_0) (* yi (sin t_0))))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = 2.0f * (uy * ((float) M_PI));
float tmp;
if (uy <= 0.0002800000074785203f) {
tmp = fmaf(2.0f, ((uy * (yi * ((float) M_PI))) * sqrtf((1.0f - ((maxCos * maxCos) * powf((ux * (1.0f - ux)), 2.0f))))), (xi * sqrtf((1.0f + ((maxCos * maxCos) * ((ux * ux) * ((1.0f - ux) * (ux - 1.0f)))))))) + ((zi * (maxCos * (1.0f - ux))) * ux);
} else {
tmp = fmaf(xi, cosf(t_0), (yi * sinf(t_0)));
}
return tmp;
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(Float32(2.0) * Float32(uy * Float32(pi))) tmp = Float32(0.0) if (uy <= Float32(0.0002800000074785203)) tmp = Float32(fma(Float32(2.0), Float32(Float32(uy * Float32(yi * Float32(pi))) * sqrt(Float32(Float32(1.0) - Float32(Float32(maxCos * maxCos) * (Float32(ux * Float32(Float32(1.0) - ux)) ^ Float32(2.0)))))), Float32(xi * sqrt(Float32(Float32(1.0) + Float32(Float32(maxCos * maxCos) * Float32(Float32(ux * ux) * Float32(Float32(Float32(1.0) - ux) * Float32(ux - Float32(1.0))))))))) + Float32(Float32(zi * Float32(maxCos * Float32(Float32(1.0) - ux))) * ux)); else tmp = fma(xi, cos(t_0), Float32(yi * sin(t_0))); end return tmp end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := 2 \cdot \left(uy \cdot \pi\right)\\
\mathbf{if}\;uy \leq 0.0002800000074785203:\\
\;\;\;\;\mathsf{fma}\left(2, \left(uy \cdot \left(yi \cdot \pi\right)\right) \cdot \sqrt{1 - \left(maxCos \cdot maxCos\right) \cdot {\left(ux \cdot \left(1 - ux\right)\right)}^{2}}, xi \cdot \sqrt{1 + \left(maxCos \cdot maxCos\right) \cdot \left(\left(ux \cdot ux\right) \cdot \left(\left(1 - ux\right) \cdot \left(ux - 1\right)\right)\right)}\right) + \left(zi \cdot \left(maxCos \cdot \left(1 - ux\right)\right)\right) \cdot ux\\
\mathbf{else}:\\
\;\;\;\;\mathsf{fma}\left(xi, \cos t\_0, yi \cdot \sin t\_0\right)\\
\end{array}
\end{array}
if uy < 2.80000007e-4Initial program 99.3%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3299.3
Applied rewrites99.3%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3299.3
Applied rewrites99.3%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3299.3
Applied rewrites99.3%
Taylor expanded in uy around 0
lower-fma.f32N/A
Applied rewrites98.7%
if 2.80000007e-4 < uy Initial program 98.2%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3298.2
Applied rewrites98.2%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.2
Applied rewrites98.2%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3298.2
Applied rewrites98.2%
Taylor expanded in ux around 0
lower-fma.f32N/A
lower-cos.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f32N/A
lower-*.f32N/A
lower-sin.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f3290.3
Applied rewrites90.3%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(if (<= uy 0.010999999940395355)
(+
(fma
2.0
(*
(* uy (* yi PI))
(sqrt (- 1.0 (* (* maxCos maxCos) (pow (* ux (- 1.0 ux)) 2.0)))))
(*
xi
(sqrt
(+ 1.0 (* (* maxCos maxCos) (* (* ux ux) (* (- 1.0 ux) (- ux 1.0))))))))
(* (* zi (* maxCos (- 1.0 ux))) ux))
(fma
(*
(sqrt (fma (- ux 1.0) (* (* (* maxCos ux) maxCos) (* (- 1.0 ux) ux)) 1.0))
(cos (* PI (* 2.0 uy))))
xi
(* maxCos (* ux (* zi (- 1.0 ux)))))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float tmp;
if (uy <= 0.010999999940395355f) {
tmp = fmaf(2.0f, ((uy * (yi * ((float) M_PI))) * sqrtf((1.0f - ((maxCos * maxCos) * powf((ux * (1.0f - ux)), 2.0f))))), (xi * sqrtf((1.0f + ((maxCos * maxCos) * ((ux * ux) * ((1.0f - ux) * (ux - 1.0f)))))))) + ((zi * (maxCos * (1.0f - ux))) * ux);
} else {
tmp = fmaf((sqrtf(fmaf((ux - 1.0f), (((maxCos * ux) * maxCos) * ((1.0f - ux) * ux)), 1.0f)) * cosf((((float) M_PI) * (2.0f * uy)))), xi, (maxCos * (ux * (zi * (1.0f - ux)))));
}
return tmp;
}
function code(xi, yi, zi, ux, uy, maxCos) tmp = Float32(0.0) if (uy <= Float32(0.010999999940395355)) tmp = Float32(fma(Float32(2.0), Float32(Float32(uy * Float32(yi * Float32(pi))) * sqrt(Float32(Float32(1.0) - Float32(Float32(maxCos * maxCos) * (Float32(ux * Float32(Float32(1.0) - ux)) ^ Float32(2.0)))))), Float32(xi * sqrt(Float32(Float32(1.0) + Float32(Float32(maxCos * maxCos) * Float32(Float32(ux * ux) * Float32(Float32(Float32(1.0) - ux) * Float32(ux - Float32(1.0))))))))) + Float32(Float32(zi * Float32(maxCos * Float32(Float32(1.0) - ux))) * ux)); else tmp = fma(Float32(sqrt(fma(Float32(ux - Float32(1.0)), Float32(Float32(Float32(maxCos * ux) * maxCos) * Float32(Float32(Float32(1.0) - ux) * ux)), Float32(1.0))) * cos(Float32(Float32(pi) * Float32(Float32(2.0) * uy)))), xi, Float32(maxCos * Float32(ux * Float32(zi * Float32(Float32(1.0) - ux))))); end return tmp end
\begin{array}{l}
\\
\begin{array}{l}
\mathbf{if}\;uy \leq 0.010999999940395355:\\
\;\;\;\;\mathsf{fma}\left(2, \left(uy \cdot \left(yi \cdot \pi\right)\right) \cdot \sqrt{1 - \left(maxCos \cdot maxCos\right) \cdot {\left(ux \cdot \left(1 - ux\right)\right)}^{2}}, xi \cdot \sqrt{1 + \left(maxCos \cdot maxCos\right) \cdot \left(\left(ux \cdot ux\right) \cdot \left(\left(1 - ux\right) \cdot \left(ux - 1\right)\right)\right)}\right) + \left(zi \cdot \left(maxCos \cdot \left(1 - ux\right)\right)\right) \cdot ux\\
\mathbf{else}:\\
\;\;\;\;\mathsf{fma}\left(\sqrt{\mathsf{fma}\left(ux - 1, \left(\left(maxCos \cdot ux\right) \cdot maxCos\right) \cdot \left(\left(1 - ux\right) \cdot ux\right), 1\right)} \cdot \cos \left(\pi \cdot \left(2 \cdot uy\right)\right), xi, maxCos \cdot \left(ux \cdot \left(zi \cdot \left(1 - ux\right)\right)\right)\right)\\
\end{array}
\end{array}
if uy < 0.0109999999Initial program 99.3%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3299.3
Applied rewrites99.3%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3299.3
Applied rewrites99.3%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3299.3
Applied rewrites99.3%
Taylor expanded in uy around 0
lower-fma.f32N/A
Applied rewrites93.8%
if 0.0109999999 < uy Initial program 97.7%
lift-+.f32N/A
lift-+.f32N/A
associate-+l+N/A
Applied rewrites97.7%
lift-sin.f32N/A
lift-acos.f32N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
sin-acos-revN/A
lift-*.f32N/A
lift--.f32N/A
lift-sqrt.f3297.7
lift-*.f32N/A
pow2N/A
lower-pow.f3297.7
Applied rewrites97.7%
Applied rewrites97.7%
Taylor expanded in yi around 0
lower-*.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f3250.9
Applied rewrites50.9%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0 (* ux (* zi (- 1.0 ux)))))
(if (<= uy 0.010999999940395355)
(fma
2.0
(*
(* uy (* yi PI))
(sqrt (- 1.0 (* (* maxCos maxCos) (pow (* ux (- 1.0 ux)) 2.0)))))
(fma
maxCos
t_0
(*
xi
(sqrt
(+
1.0
(* (* maxCos maxCos) (* (* ux ux) (* (- 1.0 ux) (- ux 1.0)))))))))
(fma
(*
(sqrt
(fma (- ux 1.0) (* (* (* maxCos ux) maxCos) (* (- 1.0 ux) ux)) 1.0))
(cos (* PI (* 2.0 uy))))
xi
(* maxCos t_0)))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = ux * (zi * (1.0f - ux));
float tmp;
if (uy <= 0.010999999940395355f) {
tmp = fmaf(2.0f, ((uy * (yi * ((float) M_PI))) * sqrtf((1.0f - ((maxCos * maxCos) * powf((ux * (1.0f - ux)), 2.0f))))), fmaf(maxCos, t_0, (xi * sqrtf((1.0f + ((maxCos * maxCos) * ((ux * ux) * ((1.0f - ux) * (ux - 1.0f)))))))));
} else {
tmp = fmaf((sqrtf(fmaf((ux - 1.0f), (((maxCos * ux) * maxCos) * ((1.0f - ux) * ux)), 1.0f)) * cosf((((float) M_PI) * (2.0f * uy)))), xi, (maxCos * t_0));
}
return tmp;
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = Float32(ux * Float32(zi * Float32(Float32(1.0) - ux))) tmp = Float32(0.0) if (uy <= Float32(0.010999999940395355)) tmp = fma(Float32(2.0), Float32(Float32(uy * Float32(yi * Float32(pi))) * sqrt(Float32(Float32(1.0) - Float32(Float32(maxCos * maxCos) * (Float32(ux * Float32(Float32(1.0) - ux)) ^ Float32(2.0)))))), fma(maxCos, t_0, Float32(xi * sqrt(Float32(Float32(1.0) + Float32(Float32(maxCos * maxCos) * Float32(Float32(ux * ux) * Float32(Float32(Float32(1.0) - ux) * Float32(ux - Float32(1.0)))))))))); else tmp = fma(Float32(sqrt(fma(Float32(ux - Float32(1.0)), Float32(Float32(Float32(maxCos * ux) * maxCos) * Float32(Float32(Float32(1.0) - ux) * ux)), Float32(1.0))) * cos(Float32(Float32(pi) * Float32(Float32(2.0) * uy)))), xi, Float32(maxCos * t_0)); end return tmp end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := ux \cdot \left(zi \cdot \left(1 - ux\right)\right)\\
\mathbf{if}\;uy \leq 0.010999999940395355:\\
\;\;\;\;\mathsf{fma}\left(2, \left(uy \cdot \left(yi \cdot \pi\right)\right) \cdot \sqrt{1 - \left(maxCos \cdot maxCos\right) \cdot {\left(ux \cdot \left(1 - ux\right)\right)}^{2}}, \mathsf{fma}\left(maxCos, t\_0, xi \cdot \sqrt{1 + \left(maxCos \cdot maxCos\right) \cdot \left(\left(ux \cdot ux\right) \cdot \left(\left(1 - ux\right) \cdot \left(ux - 1\right)\right)\right)}\right)\right)\\
\mathbf{else}:\\
\;\;\;\;\mathsf{fma}\left(\sqrt{\mathsf{fma}\left(ux - 1, \left(\left(maxCos \cdot ux\right) \cdot maxCos\right) \cdot \left(\left(1 - ux\right) \cdot ux\right), 1\right)} \cdot \cos \left(\pi \cdot \left(2 \cdot uy\right)\right), xi, maxCos \cdot t\_0\right)\\
\end{array}
\end{array}
if uy < 0.0109999999Initial program 99.3%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3299.3
Applied rewrites99.3%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3299.3
Applied rewrites99.3%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3299.3
Applied rewrites99.3%
Taylor expanded in uy around 0
Applied rewrites93.9%
if 0.0109999999 < uy Initial program 97.7%
lift-+.f32N/A
lift-+.f32N/A
associate-+l+N/A
Applied rewrites97.7%
lift-sin.f32N/A
lift-acos.f32N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
sin-acos-revN/A
lift-*.f32N/A
lift--.f32N/A
lift-sqrt.f3297.7
lift-*.f32N/A
pow2N/A
lower-pow.f3297.7
Applied rewrites97.7%
Applied rewrites97.7%
Taylor expanded in yi around 0
lower-*.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f3250.9
Applied rewrites50.9%
(FPCore (xi yi zi ux uy maxCos) :precision binary32 (fma (* (sqrt (fma (- ux 1.0) (* (* (* maxCos ux) maxCos) (* (- 1.0 ux) ux)) 1.0)) (cos (* PI (* 2.0 uy)))) xi (* maxCos (* ux (* zi (- 1.0 ux))))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
return fmaf((sqrtf(fmaf((ux - 1.0f), (((maxCos * ux) * maxCos) * ((1.0f - ux) * ux)), 1.0f)) * cosf((((float) M_PI) * (2.0f * uy)))), xi, (maxCos * (ux * (zi * (1.0f - ux)))));
}
function code(xi, yi, zi, ux, uy, maxCos) return fma(Float32(sqrt(fma(Float32(ux - Float32(1.0)), Float32(Float32(Float32(maxCos * ux) * maxCos) * Float32(Float32(Float32(1.0) - ux) * ux)), Float32(1.0))) * cos(Float32(Float32(pi) * Float32(Float32(2.0) * uy)))), xi, Float32(maxCos * Float32(ux * Float32(zi * Float32(Float32(1.0) - ux))))) end
\begin{array}{l}
\\
\mathsf{fma}\left(\sqrt{\mathsf{fma}\left(ux - 1, \left(\left(maxCos \cdot ux\right) \cdot maxCos\right) \cdot \left(\left(1 - ux\right) \cdot ux\right), 1\right)} \cdot \cos \left(\pi \cdot \left(2 \cdot uy\right)\right), xi, maxCos \cdot \left(ux \cdot \left(zi \cdot \left(1 - ux\right)\right)\right)\right)
\end{array}
Initial program 98.9%
lift-+.f32N/A
lift-+.f32N/A
associate-+l+N/A
Applied rewrites99.0%
lift-sin.f32N/A
lift-acos.f32N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
sin-acos-revN/A
lift-*.f32N/A
lift--.f32N/A
lift-sqrt.f3299.0
lift-*.f32N/A
pow2N/A
lower-pow.f3299.0
Applied rewrites99.0%
Applied rewrites99.0%
Taylor expanded in yi around 0
lower-*.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f3258.5
Applied rewrites58.5%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(let* ((t_0
(sqrt
(+
1.0
(* (* maxCos maxCos) (* (* ux ux) (* (- 1.0 ux) (- ux 1.0))))))))
(if (<= uy 0.0002800000074785203)
(+ (* xi t_0) (* (* zi (* maxCos (- 1.0 ux))) ux))
(* (* xi (cos (* 2.0 (* uy PI)))) t_0))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
float t_0 = sqrtf((1.0f + ((maxCos * maxCos) * ((ux * ux) * ((1.0f - ux) * (ux - 1.0f))))));
float tmp;
if (uy <= 0.0002800000074785203f) {
tmp = (xi * t_0) + ((zi * (maxCos * (1.0f - ux))) * ux);
} else {
tmp = (xi * cosf((2.0f * (uy * ((float) M_PI))))) * t_0;
}
return tmp;
}
function code(xi, yi, zi, ux, uy, maxCos) t_0 = sqrt(Float32(Float32(1.0) + Float32(Float32(maxCos * maxCos) * Float32(Float32(ux * ux) * Float32(Float32(Float32(1.0) - ux) * Float32(ux - Float32(1.0))))))) tmp = Float32(0.0) if (uy <= Float32(0.0002800000074785203)) tmp = Float32(Float32(xi * t_0) + Float32(Float32(zi * Float32(maxCos * Float32(Float32(1.0) - ux))) * ux)); else tmp = Float32(Float32(xi * cos(Float32(Float32(2.0) * Float32(uy * Float32(pi))))) * t_0); end return tmp end
function tmp_2 = code(xi, yi, zi, ux, uy, maxCos) t_0 = sqrt((single(1.0) + ((maxCos * maxCos) * ((ux * ux) * ((single(1.0) - ux) * (ux - single(1.0))))))); tmp = single(0.0); if (uy <= single(0.0002800000074785203)) tmp = (xi * t_0) + ((zi * (maxCos * (single(1.0) - ux))) * ux); else tmp = (xi * cos((single(2.0) * (uy * single(pi))))) * t_0; end tmp_2 = tmp; end
\begin{array}{l}
\\
\begin{array}{l}
t_0 := \sqrt{1 + \left(maxCos \cdot maxCos\right) \cdot \left(\left(ux \cdot ux\right) \cdot \left(\left(1 - ux\right) \cdot \left(ux - 1\right)\right)\right)}\\
\mathbf{if}\;uy \leq 0.0002800000074785203:\\
\;\;\;\;xi \cdot t\_0 + \left(zi \cdot \left(maxCos \cdot \left(1 - ux\right)\right)\right) \cdot ux\\
\mathbf{else}:\\
\;\;\;\;\left(xi \cdot \cos \left(2 \cdot \left(uy \cdot \pi\right)\right)\right) \cdot t\_0\\
\end{array}
\end{array}
if uy < 2.80000007e-4Initial program 99.3%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3299.3
Applied rewrites99.3%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3299.3
Applied rewrites99.3%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3299.3
Applied rewrites99.3%
Taylor expanded in uy around 0
lower-*.f32N/A
lower-sqrt.f32N/A
lower-+.f32N/A
lower-*.f32N/A
unpow2N/A
lower-*.f32N/A
lower-*.f32N/A
unpow2N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f32N/A
lift--.f3261.4
Applied rewrites61.4%
if 2.80000007e-4 < uy Initial program 98.2%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3298.2
Applied rewrites98.2%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.2
Applied rewrites98.2%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3298.2
Applied rewrites98.2%
Taylor expanded in xi around inf
lower-*.f32N/A
lower-*.f32N/A
lower-cos.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f32N/A
lower-sqrt.f32N/A
Applied rewrites47.2%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(+
(*
xi
(sqrt
(+ 1.0 (* (* maxCos maxCos) (* (* ux ux) (* (- 1.0 ux) (- ux 1.0)))))))
(* (* zi (* maxCos (- 1.0 ux))) ux)))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
return (xi * sqrtf((1.0f + ((maxCos * maxCos) * ((ux * ux) * ((1.0f - ux) * (ux - 1.0f))))))) + ((zi * (maxCos * (1.0f - ux))) * ux);
}
module fmin_fmax_functions
implicit none
private
public fmax
public fmin
interface fmax
module procedure fmax88
module procedure fmax44
module procedure fmax84
module procedure fmax48
end interface
interface fmin
module procedure fmin88
module procedure fmin44
module procedure fmin84
module procedure fmin48
end interface
contains
real(8) function fmax88(x, y) result (res)
real(8), intent (in) :: x
real(8), intent (in) :: y
res = merge(y, merge(x, max(x, y), y /= y), x /= x)
end function
real(4) function fmax44(x, y) result (res)
real(4), intent (in) :: x
real(4), intent (in) :: y
res = merge(y, merge(x, max(x, y), y /= y), x /= x)
end function
real(8) function fmax84(x, y) result(res)
real(8), intent (in) :: x
real(4), intent (in) :: y
res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
end function
real(8) function fmax48(x, y) result(res)
real(4), intent (in) :: x
real(8), intent (in) :: y
res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
end function
real(8) function fmin88(x, y) result (res)
real(8), intent (in) :: x
real(8), intent (in) :: y
res = merge(y, merge(x, min(x, y), y /= y), x /= x)
end function
real(4) function fmin44(x, y) result (res)
real(4), intent (in) :: x
real(4), intent (in) :: y
res = merge(y, merge(x, min(x, y), y /= y), x /= x)
end function
real(8) function fmin84(x, y) result(res)
real(8), intent (in) :: x
real(4), intent (in) :: y
res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
end function
real(8) function fmin48(x, y) result(res)
real(4), intent (in) :: x
real(8), intent (in) :: y
res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
end function
end module
real(4) function code(xi, yi, zi, ux, uy, maxcos)
use fmin_fmax_functions
real(4), intent (in) :: xi
real(4), intent (in) :: yi
real(4), intent (in) :: zi
real(4), intent (in) :: ux
real(4), intent (in) :: uy
real(4), intent (in) :: maxcos
code = (xi * sqrt((1.0e0 + ((maxcos * maxcos) * ((ux * ux) * ((1.0e0 - ux) * (ux - 1.0e0))))))) + ((zi * (maxcos * (1.0e0 - ux))) * ux)
end function
function code(xi, yi, zi, ux, uy, maxCos) return Float32(Float32(xi * sqrt(Float32(Float32(1.0) + Float32(Float32(maxCos * maxCos) * Float32(Float32(ux * ux) * Float32(Float32(Float32(1.0) - ux) * Float32(ux - Float32(1.0)))))))) + Float32(Float32(zi * Float32(maxCos * Float32(Float32(1.0) - ux))) * ux)) end
function tmp = code(xi, yi, zi, ux, uy, maxCos) tmp = (xi * sqrt((single(1.0) + ((maxCos * maxCos) * ((ux * ux) * ((single(1.0) - ux) * (ux - single(1.0)))))))) + ((zi * (maxCos * (single(1.0) - ux))) * ux); end
\begin{array}{l}
\\
xi \cdot \sqrt{1 + \left(maxCos \cdot maxCos\right) \cdot \left(\left(ux \cdot ux\right) \cdot \left(\left(1 - ux\right) \cdot \left(ux - 1\right)\right)\right)} + \left(zi \cdot \left(maxCos \cdot \left(1 - ux\right)\right)\right) \cdot ux
\end{array}
Initial program 98.9%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3298.9
Applied rewrites98.9%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.9
Applied rewrites98.9%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3298.9
Applied rewrites98.9%
Taylor expanded in uy around 0
lower-*.f32N/A
lower-sqrt.f32N/A
lower-+.f32N/A
lower-*.f32N/A
unpow2N/A
lower-*.f32N/A
lower-*.f32N/A
unpow2N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f32N/A
lift--.f3250.8
Applied rewrites50.8%
(FPCore (xi yi zi ux uy maxCos)
:precision binary32
(fma
maxCos
(* ux (* zi (- 1.0 ux)))
(*
xi
(sqrt
(+ 1.0 (* (* maxCos maxCos) (* (* ux ux) (* (- 1.0 ux) (- ux 1.0)))))))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
return fmaf(maxCos, (ux * (zi * (1.0f - ux))), (xi * sqrtf((1.0f + ((maxCos * maxCos) * ((ux * ux) * ((1.0f - ux) * (ux - 1.0f))))))));
}
function code(xi, yi, zi, ux, uy, maxCos) return fma(maxCos, Float32(ux * Float32(zi * Float32(Float32(1.0) - ux))), Float32(xi * sqrt(Float32(Float32(1.0) + Float32(Float32(maxCos * maxCos) * Float32(Float32(ux * ux) * Float32(Float32(Float32(1.0) - ux) * Float32(ux - Float32(1.0))))))))) end
\begin{array}{l}
\\
\mathsf{fma}\left(maxCos, ux \cdot \left(zi \cdot \left(1 - ux\right)\right), xi \cdot \sqrt{1 + \left(maxCos \cdot maxCos\right) \cdot \left(\left(ux \cdot ux\right) \cdot \left(\left(1 - ux\right) \cdot \left(ux - 1\right)\right)\right)}\right)
\end{array}
Initial program 98.9%
lift-*.f32N/A
*-commutativeN/A
count-2-revN/A
lower-+.f3298.9
Applied rewrites98.9%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.9
Applied rewrites98.9%
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
associate-*r*N/A
lift-*.f32N/A
*-commutativeN/A
lift-*.f32N/A
lift-*.f32N/A
lower-*.f3298.9
Applied rewrites98.9%
Taylor expanded in uy around 0
lower-fma.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f32N/A
lower-*.f32N/A
lower-sqrt.f32N/A
lower-+.f32N/A
Applied rewrites50.8%
(FPCore (xi yi zi ux uy maxCos) :precision binary32 (* maxCos (* ux (* zi (- 1.0 ux)))))
float code(float xi, float yi, float zi, float ux, float uy, float maxCos) {
return maxCos * (ux * (zi * (1.0f - ux)));
}
module fmin_fmax_functions
implicit none
private
public fmax
public fmin
interface fmax
module procedure fmax88
module procedure fmax44
module procedure fmax84
module procedure fmax48
end interface
interface fmin
module procedure fmin88
module procedure fmin44
module procedure fmin84
module procedure fmin48
end interface
contains
real(8) function fmax88(x, y) result (res)
real(8), intent (in) :: x
real(8), intent (in) :: y
res = merge(y, merge(x, max(x, y), y /= y), x /= x)
end function
real(4) function fmax44(x, y) result (res)
real(4), intent (in) :: x
real(4), intent (in) :: y
res = merge(y, merge(x, max(x, y), y /= y), x /= x)
end function
real(8) function fmax84(x, y) result(res)
real(8), intent (in) :: x
real(4), intent (in) :: y
res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
end function
real(8) function fmax48(x, y) result(res)
real(4), intent (in) :: x
real(8), intent (in) :: y
res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
end function
real(8) function fmin88(x, y) result (res)
real(8), intent (in) :: x
real(8), intent (in) :: y
res = merge(y, merge(x, min(x, y), y /= y), x /= x)
end function
real(4) function fmin44(x, y) result (res)
real(4), intent (in) :: x
real(4), intent (in) :: y
res = merge(y, merge(x, min(x, y), y /= y), x /= x)
end function
real(8) function fmin84(x, y) result(res)
real(8), intent (in) :: x
real(4), intent (in) :: y
res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
end function
real(8) function fmin48(x, y) result(res)
real(4), intent (in) :: x
real(8), intent (in) :: y
res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
end function
end module
real(4) function code(xi, yi, zi, ux, uy, maxcos)
use fmin_fmax_functions
real(4), intent (in) :: xi
real(4), intent (in) :: yi
real(4), intent (in) :: zi
real(4), intent (in) :: ux
real(4), intent (in) :: uy
real(4), intent (in) :: maxcos
code = maxcos * (ux * (zi * (1.0e0 - ux)))
end function
function code(xi, yi, zi, ux, uy, maxCos) return Float32(maxCos * Float32(ux * Float32(zi * Float32(Float32(1.0) - ux)))) end
function tmp = code(xi, yi, zi, ux, uy, maxCos) tmp = maxCos * (ux * (zi * (single(1.0) - ux))); end
\begin{array}{l}
\\
maxCos \cdot \left(ux \cdot \left(zi \cdot \left(1 - ux\right)\right)\right)
\end{array}
Initial program 98.9%
lift-cos.f32N/A
sin-+PI/2-revN/A
lower-sin.f32N/A
lift-*.f32N/A
*-commutativeN/A
lower-fma.f32N/A
lift-*.f32N/A
*-commutativeN/A
lower-*.f32N/A
lift-PI.f32N/A
lower-/.f3298.9
Applied rewrites98.9%
lift--.f32N/A
sub-negate1N/A
+-commutativeN/A
lift-*.f32N/A
distribute-lft-neg-inN/A
lower-fma.f32N/A
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
distribute-lft-neg-inN/A
lift--.f32N/A
sub-negate2N/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f3298.9
Applied rewrites98.9%
Taylor expanded in zi around inf
lower-*.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift--.f3213.3
Applied rewrites13.3%
herbie shell --seed 2025107
(FPCore (xi yi zi ux uy maxCos)
:name "UniformSampleCone 2"
:precision binary32
:pre (and (and (and (and (and (and (<= -10000.0 xi) (<= xi 10000.0)) (and (<= -10000.0 yi) (<= yi 10000.0))) (and (<= -10000.0 zi) (<= zi 10000.0))) (and (<= 2.328306437e-10 ux) (<= ux 1.0))) (and (<= 2.328306437e-10 uy) (<= uy 1.0))) (and (<= 0.0 maxCos) (<= maxCos 1.0)))
(+ (+ (* (* (cos (* (* uy 2.0) PI)) (sqrt (- 1.0 (* (* (* (- 1.0 ux) maxCos) ux) (* (* (- 1.0 ux) maxCos) ux))))) xi) (* (* (sin (* (* uy 2.0) PI)) (sqrt (- 1.0 (* (* (* (- 1.0 ux) maxCos) ux) (* (* (- 1.0 ux) maxCos) ux))))) yi)) (* (* (* (- 1.0 ux) maxCos) ux) zi)))