tree_d_3d.kernels.fmm_traverse
The fmm_traverse kernel of the Barnes-Hut and FMM solvers: Pyccel, NumPy and CUDA versions with the same arguments.
kernel(...) runs the CUDA version when the arrays are CuPy arrays, and otherwise the
Pyccel version, or the NumPy version if Pyccel is unavailable (see
cunumpy.kernels.set_host_kernel_implementation to choose).
Attributes
| Name | Description |
|---|---|
kernel | No description. |
Modules
fmm_traverse_metalThe Metal version (Apple-silicon GPU, float32) of the fmm_traverse kernel.fmm_traverse_numpyThe NumPy version of the fmm_traverse kernel, for machines without Pyccel.fmm_traverse_pyccelThe Pyccel version (CPU) of the fmm_traverse kernel of the Barnes-Hut and FMM solvers.
kernelattributemodule attribute#
kernel = xp.kernels.Kernel.from_folder(__name__, **KERNEL_OPTIONS)