Skip to main content

Accelerator Scenarios

This chapter covers end-to-end inference scenarios on domestic / heterogeneous accelerators (such as Huawei Ascend and Hygon). For general inference image, template, and deployment operations, see AI Inference.

Scenario Overview

ScenarioDescription
Run DeepSeek V4 Flash on Huawei Ascend NPUDeploy DeepSeek-V4-Flash with vLLM Ascend on Huawei Ascend NPU
Hygon GPUPlatform support is available; for general inference operations, see AI Inference