Help Center> ModelArts> Best Practices> Ascend Application Instructions> Application Scenarios
View PDF

Application Scenarios

Ascend 310 and Ascend 910 are Huawei-developed cloud AI chips. Ascend 310 focuses on ultra-high computing efficiency and low power consumption. Based on the features of low power consumption and high computing power, ModelArts supports Ascend 310 for high-performance inference capabilities. Ascend 910 is a training chip with higher performance. It is powerful and small. ModelArts supports Ascend 910 for training capabilities and provides algorithms developed using the MindSpore engine.

ModelArts provides built-in algorithms (in Training Management or AI Market). You can use Ascend 910 to train models and Ascend 310 to deploy services. There are two types of algorithms: built-in algorithms and cloud algorithms released in AI Market. Algorithms That Support Ascend Chips and Their Application Scenarios lists the algorithms that support the Ascend chips.

Algorithms That Support Ascend Chips and Their Application Scenarios

ModelArts built-in algorithms that support object detection, image classification, text classification, and sound classification.

In the same scenario, different algorithms support different training resources and inference resources. Read the algorithm description carefully and select the corresponding algorithm and resource specifications to build a model. For example, if you want to use Ascend 910 for training and deploy the model on Ascend 310, you are advised to select an algorithm that supports both Ascend 910 and Ascend 310, for example, ResNet50 (image classification/MindSpore/Ascend910 training) or ResNet_v1_50 (image classification/TensorFlow/Ascend910 training).

Ascend Application Samples

This document provides the following samples for algorithms that support Ascend applications. You can use built-in algorithms provided by ModelArts to support your service applications by referring to the following samples.

Table 1 Ascend samples

Sample

Engine

Function

Scenario

Description

Using a Built-in Algorithm for Object Detection (Ascend 310)

TensorFlow

Built-in algorithms (in Training Management) and Ascend 310 inference

Object detection

Use a built-in algorithm to train a model, and use Ascend 310 to deploy the model as a real-time service.