Python Tutorial Node: Attributes With CUDA Array Pointers In Cpu Memory

This is a tutorial node. It illustrates the alternative method of extracting pointers to GPU array data in which the pointer returned is a CPU pointer and can be dereferenced on the CPU side. Without the cudaPointers value set that pointer would be a GPU pointer to an array of GPU pointers and could only be dereferenced on the device. Both plain attribute and bundle attribute extraction are shown.

Installation

To use this Node, you must enable omni.graph.tutorials in the Extension Manager.

Inputs

Name	Type	Description	Default
points	float[3][]	Array of points to be moved	[]
multiplier	float[3]	Amplitude of the expansion for the input points	[1.0, 1.0, 1.0]

Outputs

Name	Type	Description	Default
points	float[3][]	Final positions of points
outBundle	bundle	Bundle containing a copy of the output points