The nvmath.device module is experimental and potentially subject to future changes.

cuFFTDx APIs (nvmath.device)#

Overview#

These APIs offer integration with the NVIDIA cuFFTDx library. Detailed documentation of cuFFTDx can be found in the cuFFTDx documentation.

Note

The FFT device APIs support both the numba-cuda and the numba-cuda-mlir compilers. See Supported Compilers for details.

Note

The FFT device APIs in module nvmath.device currently support cuFFTDx 1.7.0, also available as part of MathDx 26.03.0. All functionalities from the C++ library are supported with the exception of cuFFTDx C++ APIs with a workspace argument, which are currently not available in nvmath-python. This means that certain FFT sizes and configurations that require a workspace are not supported. For more information on which sizes and configurations require a workspace, please refer to the cuFFTDx Supported Functionality documentation.

API Reference#

FFT(size, precision, fft_type, execution, *)

A class that encapsulates a partial FFT device function.
On this page