Prerequisites
Prepare the cluster and nodes before installing Alauda Build of NPU Operator.
TOC
ACP and cluster accessNode and hardwareDriver and runtimeDevice exposure modelOptional componentsACP cross referencesACP and cluster access
- ACP v4.0-v4.4.
- Cluster administrator access to the target workload cluster where
Alauda Build of NPU Operatorwill be installed. - OperatorHub and Cluster Plugin installation are available in ACP.
- Package download from Customer Portal and package upload in ACP are available.
Node and hardware
- Ascend 910B or Ascend 310P worker nodes.
- ARM architecture.
Alauda Build of Node Feature Discoveryis installed on the target workload cluster so shared kernel, OS, architecture, and hardware labels are available before NPU components are reconciled.- Nodes selected for NPU components allow NPU Feature Discovery to run. The Driver, runtime, Device Plugin, and Exporter select NPU nodes through the discovered
openfuyao.com/npu.presentlabel. The optional Operator manager and some MindCluster components also usemasterselector=dls-master-nodeorworkerselector=dls-worker-node; the bundled NPU Feature Discovery maintains those role labels when it runs. Do not add them manually as a normal installation step. - For Ascend 910-family nodes, complete HCCN device IP configuration before workload verification. v26.6.0 does not initialize HCCN device IPs automatically.
Driver and runtime
Choose one driver management scenario before creating the NPUOperatorCtl instance:
In scenario 1, verify that /usr/local/Ascend/driver exists on every selected
NPU node before enabling NPU components that depend on the driver.
For scenario 2.2, import the matching ARM64 Driver image separately from the product package. Map the complete source image reference to the target workload cluster registry and make sure the admitted image is allowed by ImageWhiteList when the cluster enforces image policies.
Device exposure model
Use Choose an Ascend NPU Allocation Mode to evaluate performance, device utilization, sharing requirements, and operational complexity. Then choose one Ascend device exposure owner for the complete scope managed by each NPUOperatorCtl instance:
Do not run multiple device exposure owners for the same Ascend devices in one managed scope unless the current product versions explicitly document and validate mutually exclusive Device Plugin placement.
Optional components
- Install Volcano only if you plan to enable ClusterD.
- Pre-stage MindIO SDK zip files on each target NPU node only when you plan to enable MindIO TFT or MindIO ACP in an offline or controlled network environment.
ACP cross references
- For package upload, see Upload Packages.
- For quota resource display, see Accelerator Resource Quota.
- For dashboard management, see Manage dashboards.