QueueData

English | 简体中文

Product Support Status

Product	Supported
Ascend 950PR&950DT products	√
Atlas A3 products	√
Atlas A2 products	√
Atlas 200I/500 A2 inference products	√
Atlas inference products	√
Atlas training products	√

Function Description

Provides queue data for other operators. The QueueData operator serves as a queue data input node in a graph. It computes the length of the output tensor based on the output_types and output_shapes attributes (each item contains a 64-byte information header, dimension description, and data bytes), and outputs a one-dimensional tensor of the DT_UINT8 type.

Parameter Description

Parameter Name	Input/Output/Attribute	Description	Data Type	Data Format
index	Attribute	Index of the input tensor. The data type is int32 or int64. Assume that a network has three data nodes: one should be set to 0, another to 1, and the third to 2.	Int, with the default value 0	-
queue_name	Attribute	Queue name.	String, with the default value ""	-
output_types	Attribute	List of data types of the output data.	ListType, with the default value {}	-
output_shapes	Attribute	List of shapes of the output data.	ListListInt, with the default value {{}, {}}	-
y	Output	One-dimensional tensor of the DT_UINT8 type, whose length is the total number of bytes computed based on output_types and output_shapes.	DT_UINT8	ND

Constraint Description

• The lengths of output_types and output_shapes must be consistent.

Invocation Description

The QueueData operator is a basic operator of the graph engine. It is usually used as a queue data input node in a graph and is not invoked separately.

Invocation Method	Invocation Sample	Description
Graph-mode invocation	test_geir_queue_data	Invokes the QueueData operator by building a graph through the operator IR