OpenCL default kernels
This section lists all kernel functions that are available to the
opencl
filter if no filename is specified.
-
void fix_nan_and_inf()
Sets element to 0.0 if it is NaN or Inf.
-
void absorptivity()
Computes \(f(x) = -log(x)\).
-
void nlm_noise_reduction()
Smooths data within a local neighbourhood.
-
void diff()
Computes \(f(x, y) = x - y\).
OpenCL reduction default kernels
This section lists all kernel functions that are available to the
opencl-reduce
filter if no filename is specified. These kernels
are supposed to be used for the kernel
argument.
-
void minimum()
Computes the minimum of each pixel in the stream.
-
void maximum()
Computes the maximum of each pixel in the stream.
-
void sum()
Computes the sum of each pixel in the stream.
These kernels are supposed to be used in the finish
argument:
-
void divide()
Divides each pixel by the stream count. Together with
sum
this can be used to compute the average, i.e.:ufo-launch .. ! opencl-reduce kernel=sum finish=divide ! ..