Single-Operator API Execution

This section describes the APIs (such as aclnn and acldvpp) for calling operators in single-operator API execution mode. These APIs are important components of the CANN operator library (Operator Library API Reference). Table 1 lists the APIs.

The functions, prototypes, parameters, and calling examples in this section help you understand how to use operators and develop AI services efficiently.

Table 1 Operator APIs

API Category

Description

API Documentation

Operator APIs (aclnn) > Math APIs

Mathematical calculation operator library, which provides operators like Add and Abs. These APIs are prefixed with aclnnXxx.

See in Operator Library API Reference.

Operator APIs (aclnn) > NN APIs

Neural network (NN) operator library, which provides operators like Matmul. These APIs are prefixed with aclnnXxx.

Currently, this type of operators accounts for the largest proportion in the operator library.

Operator APIs (aclnn) > CV APIs

Computer vision (CV) operator library, which provides operators like GridSample. These APIs are prefixed with aclnnXxx.

Operator APIs (aclnn) > Transformer APIs

Large model computing operator library, which provides operators like FlashAttention, MC2 (merged compute and communication), and mixture of experts (MoE).

These APIs are prefixed with aclnnXxx.

Media data processing APIs (acldvpp)

Digital vision preprocessing operators, which provide preprocessing APIs for high-performance video/image encoding and decoding, as well as image cropping and resizing. These APIs are prefixed with acldvppXxx.

See "Media Data Processing APIs (acldvpp)" in Operator Library API Reference.