Software Package Preparation

Software Package Download

Click here. In the select resource area on the left, filter the required software packages, confirm the version information, and obtain the software packages listed in Table 1. Based on your service requirements, install one out of Toolkit, NNAE, and NNRT in CANN packages (Kernels and NNAL packages are optional).

Once you download software, you agree to the terms and conditions of Huawei Enterprise End User License Agreement (EULA).

**Table 1** Software Package List
Software Type	Description	Name
Ascend NPU driver	The driver is deployed on an Ascend AI Processor. It manages and queries the Ascend AI Processor and provides processor control and resource allocation interfaces for the upper-layer CANN software.	Ascend-hdk-<chip_type>-npu-driver_<version>_linux-<arch>.run
Ascend NPU firmware	The firmware contains the OS, power component, and power consumption management control software of an Ascend AI Processor. It is used for model calculation, processor startup control, and power consumption control that are loaded to the Ascend AI Processor.	Ascend-hdk-<chip_type>-npu-firmware_<version>.run
Toolkit	CANN development kit, which is installed in the training, inference, and development and debugging scenarios. It is used for training and inference services, model conversion, and operator/application/model development and build.	Ascend-cann-toolkit_<version>_linux-<arch>.run
NNAE (Ascend Neural Network Acceleration Engine)	CANN deep learning engine package, which is installed in the training and inference scenarios and is used for training and inference services.	Ascend-cann-nnae_<version>_linux-<arch>.run
NNRT (Ascend Neural Network Runtime)	CANN offline inference engine package, which is installed in the edge inference scenario and supports only offline inference. It is mainly used for model inference of applications.	Ascend-cann-nnrt_<version>_linux-<arch>.run
Kernels	CANN binary operator package, including the dynamic and static library files for single-operator API execution (such as aclnn APIs) and kernel binary file. Usage Scenarios: This software package must be installed when single-operator APIs (such as aclnn APIs) are executed. In the dynamic shape scenario in graph mode, you are advised to install this software package to improve the compilation performance. In the static shape scenario in graph mode, if this software package is installed, the compilation performance can be improved. If this package is not installed, build a new kernel based on the specified shape to improve the operator execution performance. Before the installation, ensure that the Toolkit, NNRT, or NNAE software package has been installed. Select the Kernels package corresponding to the product type and architecture of the device.	Ascend-cann-kernels-<chip_type>_<version>_linux-<arch>.run
NNAL (Ascend Neural Network Acceleration Library)	CANN neural network acceleration library, including the Ascend Transformer Boost (ATB) acceleration library for large models, can improve the training and inference performance of large models. Before the installation, ensure that the Toolkit or NNAE software package has been installed.	Ascend-cann-nnal_<version>_linux-<arch>.run

<version> indicates the software version.
<arch> indicates the CPU architecture.
<chip_type> indicates the processor type.

Verifying the Digital Signature

To prevent a software package from being maliciously tampered with during transfer or storage, download also the corresponding digital signature file for integrity verification while downloading the software package.

Click PGP Verify to obtain the tool package, decompress it and verify the PGP digital signature. For details, see OpenPGP Signature Verification Guide. If the verification fails, do not use the software package. See Support to get help from the community or submit a service ticket.