# Verification

After installing the SDKs, verify that your Cloud AI hardware and software are functioning correctly.

## Quick verification checklist

| Step | Action | Expected Result |
| --- | --- | --- |
| 1 | [Verify PCIe enumeration](https://docs.qualcomm.com/doc/80-99100-3/topic/verification.html#verify-pcie) | Qualcomm devices listed |
| 2 | [Check device nodes](https://docs.qualcomm.com/doc/80-99100-3/topic/verification.html#verify-device-nodes) | `/dev/accel/accel*` files exist |
| 3 | [Check card health](https://docs.qualcomm.com/doc/80-99100-3/topic/verification.html#verify-card-health) | Status: Ready |
| 4 | [Run test inference](https://docs.qualcomm.com/doc/80-99100-3/topic/verification.html#verify-inference) | Positive Inf/Sec value |

## Verify PCIe enumeration

lspci | grep Qualcomm
    Copy to clipboard

**Expected output:**

03:00.0 Processing accelerators: Qualcomm Device a100
    Copy to clipboard

Note

If no devices appear, try:

- Re-seating the card
- Testing with a different PCIe slot
- Verifying PCIe power connections

## Check device nodes

ls /dev/accel/accel*
    Copy to clipboard

**Expected output:**

/dev/accel/accel0  /dev/accel/accel1  /dev/accel/accel2  /dev/accel/accel3
    Copy to clipboard

One device node is created for each Cloud AI SoC.

Note

If device nodes are missing, verify [MSI settings](https://docs.qualcomm.com/doc/80-99100-3/topic/prerequisites.html#reference-to-msi) in your BIOS.

## Check card health

/opt/qti-aic/tools/qaic-util -q | grep -e Status -e QID
    Copy to clipboard

**Expected output:**

QID 0
      Status:Ready
    QID 1
      Status:Ready
    QID 2
      Status:Ready
    QID 3
      Status:Ready
    Copy to clipboard

**Status meanings:**

| Status | Description |
| --- | --- |
| `Ready` | Card is healthy and operational |
| `Error` | Card hasn’t booted completely |

Tip

If status shows `Error`:

- Wait a few minutes and retry (card may still be booting)
- Check for unsupported OS/platform
- Verify secure boot settings

## Run test inference

Run a test workload to verify hardware and software integration:

/opt/qti-aic/exec/qaic-runner \
      -t /opt/qti-aic/test-data/aic100/v2/1nsp/1nsp-conv-hmx/ \
      -a 14 \
      -n 5000 \
      -d 0
    Copy to clipboard

**Command options:**

- `-t` - Path to test data
- `-a` - Number of NSPs in the target QID
- `-n` - Number of iterations
- `-d` - QID to run the workload on

**Expected output:**

Input file: /opt/qti-aic/test-data/aic100/v2/1nsp/1nsp-conv-hmx/user_idx0_in.bin
    ---- Stats ----
    InferenceCnt 5000 TotalDuration 15094us BatchSize 1 Inf/Sec 331257.453
    Copy to clipboard

A positive `Inf/Sec` value confirms that hardware and software are functioning correctly.

## Troubleshooting

**No Qualcomm devices in lspci**

- Reseat the PCIe card
- Try a different PCIe slot
- Check power connections to the card
- Verify card compatibility with your motherboard

**Missing /dev/accel nodes**

- Enable MSI in BIOS (Intel VT-d or AMD IOMMU)
- Verify 32 MSIs are available (see [MSI configuration](https://docs.qualcomm.com/doc/80-99100-3/topic/prerequisites.html#reference-to-msi))
- Reinstall Platform SDK

**Status: Error on card**

- Allow more time for boot (up to 5 minutes)
- Check kernel compatibility
- Review system logs: `dmesg | grep qaic`
- Verify secure boot configuration

**Low Inf/Sec in test**

- Adjust `-a` parameter to match your card’s NSP count
- Check for thermal throttling
- Verify PCIe link speed: `lspci -vvv -s <device> | grep LnkSta`

## Next steps

Your Cloud AI system is ready. Continue to:

- [Inference Workflow](https://docs.qualcomm.com/doc/80-99100-3/topic/index_Inference-Workflow.html) - Compile and run your models
- [System Management](https://docs.qualcomm.com/doc/80-99100-3/topic/index_system-management_System-Management.html) - Monitor device health and performance

Last Published: Aug 25, 2026

[Previous Topic
Install Apps SDK](https://docs.qualcomm.com/bundle/publicresource/80-99100-3/topics/apps-sdk.md) [Next Topic
Model inference and deployment](https://docs.qualcomm.com/bundle/publicresource/80-99100-3/topics/index_Inference-Workflow.md)