
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (sqrt (- (log (- 1.0 u1)))) (cos (* (* 2.0 (PI)) u2))))
\begin{array}{l}
\\
\sqrt{-\log \left(1 - u1\right)} \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot u2\right)
\end{array}
Sampling outcomes in binary32 precision:
Herbie found 12 alternatives:
| Alternative | Accuracy | Speedup |
|---|
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (sqrt (- (log (- 1.0 u1)))) (cos (* (* 2.0 (PI)) u2))))
\begin{array}{l}
\\
\sqrt{-\log \left(1 - u1\right)} \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot u2\right)
\end{array}
(FPCore (cosTheta_i u1 u2)
:precision binary32
(let* ((t_0 (cbrt (PI))))
(if (<= u1 0.03500000014901161)
(*
(sqrt
(-
(*
(- (* (- (* (- (* -0.25 u1) 0.3333333333333333) u1) 0.5) u1) 1.0)
u1)))
(cos (* (* 2.0 (PI)) u2)))
(*
(sqrt (- (log (- 1.0 u1))))
(cos (* (pow t_0 2.0) (* (* t_0 u2) 2.0)))))))\begin{array}{l}
\\
\begin{array}{l}
t_0 := \sqrt[3]{\mathsf{PI}\left(\right)}\\
\mathbf{if}\;u1 \leq 0.03500000014901161:\\
\;\;\;\;\sqrt{-\left(\left(\left(-0.25 \cdot u1 - 0.3333333333333333\right) \cdot u1 - 0.5\right) \cdot u1 - 1\right) \cdot u1} \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot u2\right)\\
\mathbf{else}:\\
\;\;\;\;\sqrt{-\log \left(1 - u1\right)} \cdot \cos \left({t\_0}^{2} \cdot \left(\left(t\_0 \cdot u2\right) \cdot 2\right)\right)\\
\end{array}
\end{array}
if u1 < 0.0350000001Initial program 53.1%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
metadata-evalN/A
distribute-lft-neg-inN/A
lower--.f32N/A
distribute-lft-neg-inN/A
metadata-evalN/A
lower-*.f3299.0
Applied rewrites99.0%
if 0.0350000001 < u1 Initial program 98.0%
lift-*.f32N/A
lift-*.f32N/A
associate-*l*N/A
*-commutativeN/A
lift-PI.f32N/A
add-cube-cbrtN/A
associate-*l*N/A
associate-*l*N/A
lower-*.f32N/A
pow2N/A
lower-pow.f32N/A
lift-PI.f32N/A
lower-cbrt.f32N/A
lower-*.f32N/A
lower-*.f32N/A
lift-PI.f32N/A
lower-cbrt.f3298.1
Applied rewrites98.1%
(FPCore (cosTheta_i u1 u2)
:precision binary32
(let* ((t_0 (cos (* (* 2.0 (PI)) u2))))
(if (<= u1 0.04500000178813934)
(*
(sqrt
(-
(*
(- (* (- (* (- (* -0.25 u1) 0.3333333333333333) u1) 0.5) u1) 1.0)
u1)))
t_0)
(* (sqrt (- (log (- 1.0 u1)))) t_0))))\begin{array}{l}
\\
\begin{array}{l}
t_0 := \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot u2\right)\\
\mathbf{if}\;u1 \leq 0.04500000178813934:\\
\;\;\;\;\sqrt{-\left(\left(\left(-0.25 \cdot u1 - 0.3333333333333333\right) \cdot u1 - 0.5\right) \cdot u1 - 1\right) \cdot u1} \cdot t\_0\\
\mathbf{else}:\\
\;\;\;\;\sqrt{-\log \left(1 - u1\right)} \cdot t\_0\\
\end{array}
\end{array}
if u1 < 0.0450000018Initial program 53.5%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
metadata-evalN/A
distribute-lft-neg-inN/A
lower--.f32N/A
distribute-lft-neg-inN/A
metadata-evalN/A
lower-*.f3299.0
Applied rewrites99.0%
if 0.0450000018 < u1 Initial program 98.3%
(FPCore (cosTheta_i u1 u2)
:precision binary32
(*
(sqrt
(-
(* (- (* (- (* (- (* -0.25 u1) 0.3333333333333333) u1) 0.5) u1) 1.0) u1)))
(cos (* (* 2.0 (PI)) u2))))\begin{array}{l}
\\
\sqrt{-\left(\left(\left(-0.25 \cdot u1 - 0.3333333333333333\right) \cdot u1 - 0.5\right) \cdot u1 - 1\right) \cdot u1} \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot u2\right)
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
metadata-evalN/A
distribute-lft-neg-inN/A
lower--.f32N/A
distribute-lft-neg-inN/A
metadata-evalN/A
lower-*.f3294.0
Applied rewrites94.0%
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (sqrt (- (* (- (* (- (* -0.3333333333333333 u1) 0.5) u1) 1.0) u1))) (cos (* (* 2.0 (PI)) u2))))
\begin{array}{l}
\\
\sqrt{-\left(\left(-0.3333333333333333 \cdot u1 - 0.5\right) \cdot u1 - 1\right) \cdot u1} \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot u2\right)
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
metadata-evalN/A
distribute-lft-neg-inN/A
*-commutativeN/A
distribute-lft-neg-inN/A
lower--.f32N/A
distribute-lft-neg-inN/A
*-commutativeN/A
distribute-lft-neg-inN/A
metadata-evalN/A
lower-*.f3292.4
Applied rewrites92.4%
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (sqrt (+ (* (* 0.5 u1) u1) u1)) (cos (* (* 2.0 (PI)) u2))))
\begin{array}{l}
\\
\sqrt{\left(0.5 \cdot u1\right) \cdot u1 + u1} \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot u2\right)
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
+-commutativeN/A
lower-fma.f3258.0
Applied rewrites58.3%
Applied rewrites88.7%
(FPCore (cosTheta_i u1 u2)
:precision binary32
(if (<= u2 0.0008200000156648457)
(*
(sqrt
(-
(*
(- (* (- (* (- (* -0.25 u1) 0.3333333333333333) u1) 0.5) u1) 1.0)
u1)))
1.0)
(* (sqrt u1) (cos (* (* 2.0 (PI)) u2)))))\begin{array}{l}
\\
\begin{array}{l}
\mathbf{if}\;u2 \leq 0.0008200000156648457:\\
\;\;\;\;\sqrt{-\left(\left(\left(-0.25 \cdot u1 - 0.3333333333333333\right) \cdot u1 - 0.5\right) \cdot u1 - 1\right) \cdot u1} \cdot 1\\
\mathbf{else}:\\
\;\;\;\;\sqrt{u1} \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot u2\right)\\
\end{array}
\end{array}
if u2 < 8.20000016e-4Initial program 59.7%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
metadata-evalN/A
distribute-lft-neg-inN/A
lower--.f32N/A
distribute-lft-neg-inN/A
metadata-evalN/A
lower-*.f3294.5
Applied rewrites94.5%
Taylor expanded in u2 around 0
Applied rewrites93.3%
if 8.20000016e-4 < u2 Initial program 58.3%
lift-log.f32N/A
lift--.f32N/A
flip--N/A
log-divN/A
lower--.f32N/A
metadata-evalN/A
fp-cancel-sub-sign-invN/A
lower-log1p.f32N/A
lower-*.f32N/A
lower-neg.f32N/A
lower-log1p.f3210.1
Applied rewrites10.2%
Taylor expanded in u1 around 0
lower-sqrt.f3274.9
Applied rewrites74.9%
(FPCore (cosTheta_i u1 u2)
:precision binary32
(*
(sqrt
(-
(* (- (* (- (* (- (* -0.25 u1) 0.3333333333333333) u1) 0.5) u1) 1.0) u1)))
1.0))
float code(float cosTheta_i, float u1, float u2) {
return sqrtf(-(((((((-0.25f * u1) - 0.3333333333333333f) * u1) - 0.5f) * u1) - 1.0f) * u1)) * 1.0f;
}
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(costheta_i, u1, u2)
use fmin_fmax_functions
real(4), intent (in) :: costheta_i
real(4), intent (in) :: u1
real(4), intent (in) :: u2
code = sqrt(-((((((((-0.25e0) * u1) - 0.3333333333333333e0) * u1) - 0.5e0) * u1) - 1.0e0) * u1)) * 1.0e0
end function
function code(cosTheta_i, u1, u2) return Float32(sqrt(Float32(-Float32(Float32(Float32(Float32(Float32(Float32(Float32(Float32(-0.25) * u1) - Float32(0.3333333333333333)) * u1) - Float32(0.5)) * u1) - Float32(1.0)) * u1))) * Float32(1.0)) end
function tmp = code(cosTheta_i, u1, u2) tmp = sqrt(-(((((((single(-0.25) * u1) - single(0.3333333333333333)) * u1) - single(0.5)) * u1) - single(1.0)) * u1)) * single(1.0); end
\begin{array}{l}
\\
\sqrt{-\left(\left(\left(-0.25 \cdot u1 - 0.3333333333333333\right) \cdot u1 - 0.5\right) \cdot u1 - 1\right) \cdot u1} \cdot 1
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
metadata-evalN/A
distribute-lft-neg-inN/A
lower--.f32N/A
distribute-lft-neg-inN/A
metadata-evalN/A
lower-*.f3294.0
Applied rewrites94.0%
Taylor expanded in u2 around 0
Applied rewrites76.4%
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (sqrt (- (* (- (* (- (* -0.3333333333333333 u1) 0.5) u1) 1.0) u1))) 1.0))
float code(float cosTheta_i, float u1, float u2) {
return sqrtf(-(((((-0.3333333333333333f * u1) - 0.5f) * u1) - 1.0f) * u1)) * 1.0f;
}
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(costheta_i, u1, u2)
use fmin_fmax_functions
real(4), intent (in) :: costheta_i
real(4), intent (in) :: u1
real(4), intent (in) :: u2
code = sqrt(-((((((-0.3333333333333333e0) * u1) - 0.5e0) * u1) - 1.0e0) * u1)) * 1.0e0
end function
function code(cosTheta_i, u1, u2) return Float32(sqrt(Float32(-Float32(Float32(Float32(Float32(Float32(Float32(-0.3333333333333333) * u1) - Float32(0.5)) * u1) - Float32(1.0)) * u1))) * Float32(1.0)) end
function tmp = code(cosTheta_i, u1, u2) tmp = sqrt(-(((((single(-0.3333333333333333) * u1) - single(0.5)) * u1) - single(1.0)) * u1)) * single(1.0); end
\begin{array}{l}
\\
\sqrt{-\left(\left(-0.3333333333333333 \cdot u1 - 0.5\right) \cdot u1 - 1\right) \cdot u1} \cdot 1
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
lower--.f32N/A
*-commutativeN/A
lower-*.f32N/A
metadata-evalN/A
distribute-lft-neg-inN/A
*-commutativeN/A
distribute-lft-neg-inN/A
lower--.f32N/A
distribute-lft-neg-inN/A
*-commutativeN/A
distribute-lft-neg-inN/A
metadata-evalN/A
lower-*.f3292.4
Applied rewrites92.4%
Taylor expanded in u2 around 0
Applied rewrites75.3%
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (sqrt (+ (* (* 0.5 u1) u1) u1)) 1.0))
float code(float cosTheta_i, float u1, float u2) {
return sqrtf((((0.5f * u1) * u1) + u1)) * 1.0f;
}
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(costheta_i, u1, u2)
use fmin_fmax_functions
real(4), intent (in) :: costheta_i
real(4), intent (in) :: u1
real(4), intent (in) :: u2
code = sqrt((((0.5e0 * u1) * u1) + u1)) * 1.0e0
end function
function code(cosTheta_i, u1, u2) return Float32(sqrt(Float32(Float32(Float32(Float32(0.5) * u1) * u1) + u1)) * Float32(1.0)) end
function tmp = code(cosTheta_i, u1, u2) tmp = sqrt((((single(0.5) * u1) * u1) + u1)) * single(1.0); end
\begin{array}{l}
\\
\sqrt{\left(0.5 \cdot u1\right) \cdot u1 + u1} \cdot 1
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
+-commutativeN/A
lower-fma.f3258.4
Applied rewrites58.6%
Taylor expanded in u2 around 0
Applied rewrites65.0%
Applied rewrites73.3%
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (sqrt (* (- 1.0 (* -0.5 u1)) u1)) 1.0))
float code(float cosTheta_i, float u1, float u2) {
return sqrtf(((1.0f - (-0.5f * u1)) * u1)) * 1.0f;
}
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(costheta_i, u1, u2)
use fmin_fmax_functions
real(4), intent (in) :: costheta_i
real(4), intent (in) :: u1
real(4), intent (in) :: u2
code = sqrt(((1.0e0 - ((-0.5e0) * u1)) * u1)) * 1.0e0
end function
function code(cosTheta_i, u1, u2) return Float32(sqrt(Float32(Float32(Float32(1.0) - Float32(Float32(-0.5) * u1)) * u1)) * Float32(1.0)) end
function tmp = code(cosTheta_i, u1, u2) tmp = sqrt(((single(1.0) - (single(-0.5) * u1)) * u1)) * single(1.0); end
\begin{array}{l}
\\
\sqrt{\left(1 - -0.5 \cdot u1\right) \cdot u1} \cdot 1
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
+-commutativeN/A
lower-fma.f3259.3
Applied rewrites58.8%
Taylor expanded in u2 around 0
Applied rewrites64.7%
Applied rewrites73.2%
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (sqrt (* 1.0 u1)) 1.0))
float code(float cosTheta_i, float u1, float u2) {
return sqrtf((1.0f * u1)) * 1.0f;
}
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(costheta_i, u1, u2)
use fmin_fmax_functions
real(4), intent (in) :: costheta_i
real(4), intent (in) :: u1
real(4), intent (in) :: u2
code = sqrt((1.0e0 * u1)) * 1.0e0
end function
function code(cosTheta_i, u1, u2) return Float32(sqrt(Float32(Float32(1.0) * u1)) * Float32(1.0)) end
function tmp = code(cosTheta_i, u1, u2) tmp = sqrt((single(1.0) * u1)) * single(1.0); end
\begin{array}{l}
\\
\sqrt{1 \cdot u1} \cdot 1
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
+-commutativeN/A
lower-fma.f3258.4
Applied rewrites59.3%
Taylor expanded in u2 around 0
Applied rewrites64.7%
Taylor expanded in u1 around 0
Applied rewrites65.0%
(FPCore (cosTheta_i u1 u2) :precision binary32 (* (- (sqrt u1)) 1.0))
float code(float cosTheta_i, float u1, float u2) {
return -sqrtf(u1) * 1.0f;
}
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(costheta_i, u1, u2)
use fmin_fmax_functions
real(4), intent (in) :: costheta_i
real(4), intent (in) :: u1
real(4), intent (in) :: u2
code = -sqrt(u1) * 1.0e0
end function
function code(cosTheta_i, u1, u2) return Float32(Float32(-sqrt(u1)) * Float32(1.0)) end
function tmp = code(cosTheta_i, u1, u2) tmp = -sqrt(u1) * single(1.0); end
\begin{array}{l}
\\
\left(-\sqrt{u1}\right) \cdot 1
\end{array}
Initial program 59.2%
Taylor expanded in u1 around 0
*-commutativeN/A
lower-*.f32N/A
+-commutativeN/A
lower-fma.f3258.6
Applied rewrites58.9%
Taylor expanded in u2 around 0
Applied rewrites64.7%
Taylor expanded in u1 around 0
*-commutativeN/A
unpow2N/A
rem-square-sqrtN/A
lower-*.f32N/A
lower-sqrt.f325.2
Applied rewrites5.2%
Taylor expanded in u1 around 0
*-commutativeN/A
unpow2N/A
rem-square-sqrtN/A
mul-1-negN/A
lower-neg.f32N/A
lower-sqrt.f325.2
Applied rewrites5.2%
herbie shell --seed 2024346
(FPCore (cosTheta_i u1 u2)
:name "Beckmann Sample, near normal, slope_x"
:precision binary32
:pre (and (and (and (> cosTheta_i 0.9999) (<= cosTheta_i 1.0)) (and (<= 2.328306437e-10 u1) (<= u1 1.0))) (and (<= 2.328306437e-10 u2) (<= u2 1.0)))
(* (sqrt (- (log (- 1.0 u1)))) (cos (* (* 2.0 (PI)) u2))))