Installation

This guide walks you through installing the Cloud AI SDK on your system. Choose your installation path based on your environment:

Environment

Description

Cloud Instances

Pre-configured instances on AWS or Cirrascale (fastest setup)

Local Server

On-premise servers with Cloud AI hardware

Docker

Containerized deployment

Hypervisors

KVM, ESXi, or Hyper-V environments

Cloud Instances

Qualcomm Cloud AI accelerators are available from two cloud providers:

Note

SDKs are pre-installed on cloud instances. Skip ahead to Inference Workflow.

Local Server Installation

Follow these steps to install on an on-premise server:

  1. Verify hardware requirements - Hardware Requirements

  2. Check OS compatibility - Supported Operating Systems

  3. Configure BIOS settings - Enable MSI

  4. Download SDKs - Download from QPM

  5. Install Platform SDK - Platform SDK Installation

  6. Install Apps SDK - Apps SDK Installation

  7. Verify installation - Verification Checklist

Tip

To run Platform SDK tools without sudo, add yourself to the qaic group:

sudo usermod -aG qaic $USER
newgrp qaic  # Apply changes without logging out

Virtual Machine Installation

Cloud AI supports PCIe passthrough to virtual machines. The VM has exclusive access to the Cloud AI device (no sharing between VMs or with the host).

Supported Hypervisors (x86-64 only)

  • KVM

  • Hyper-V

  • ESXi

  • Xen

Installation Steps

  1. Configure PCIe passthrough on your hypervisor

  2. Enable MSI in BIOS and hypervisor settings

  3. Launch VM with a supported OS

  4. Install Cloud AI SDKs inside the VM

  5. Verify card health

Note

Cloud AI SDKs are installed only on the guest VM, not on the hypervisor host.

See Hypervisors for detailed PCIe passthrough and hypervisor configuration instructions.

Docker Installation

For containerized deployments:

  1. Install Platform SDK on the host system

  2. Download a prebuilt Docker image for Cloud AI or use the Apps SDK to build a custom Docker image.

  3. Launch container with device access

See Docker for detailed instructions, available images, and workflow examples.

Next Steps

After installation, proceed to: