Tutorial Node: Attributes With CUDA Data

This is a tutorial node. It performs different functions on the GPU to illustrate different types of data access. The first adds inputs ‘a’ and ‘b’ to yield output ‘sum’, all of which are on the GPU. The second is a sample expansion deformation that multiplies every point on a set of input points, stored on the GPU, by a constant value, stored on the CPU, to yield a set of output points, also on the GPU. The third is an assortment of different data types illustrating how different data is passed to the GPU. This particular node uses CUDA for its GPU computations, as indicated in the memory type value. Normal use case for GPU compute is large amounts of data. For testing purposes this node only handles a very small amount but the principle is the same.

Installation

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

Inputs

Name	Type	Description	Default
a	float	First value to be added in algorithm 1	0.0
b	float	Second value to be added in algorithm 1	0.0
points	float[3][]	Points to be moved by algorithm 2	[]
multiplier	float[3]	Amplitude of the expansion for the input points in algorithm 2	[1.0, 1.0, 1.0]
half	half	Input of type half for algorithm 3	1.0
color	colord[3]	Input with three doubles as a color for algorithm 3	[1.0, 0.5, 1.0]
matrix	matrixd[4]	Input with 16 doubles interpreted as a double-precision 4d matrix	[[1.0, 0.0, 0.0, 0.0], [0.0, 1.0, 0.0, 0.0], [0.0, 0.0, 1.0, 0.0], [0.0, 0.0, 0.0, 1.0]]

Outputs

Name	Type	Description	Default
sum	float	Sum of the two inputs from algorithm 1
points	float[3][]	Final positions of points from algorithm 2
half	half	Output of type half for algorithm 3
color	colord[3]	Output with three doubles as a color for algorithm 3
matrix	matrixd[4]	Output with 16 doubles interpreted as a double-precision 4d matrix