intel / intel/aipc-devkit-install
Feature request: add explicit post-install NPU readiness check (OpenVINO device enumeration + minimal NPU compile test)
- Dominant language
- PowerShell
- Stars
- 43
- Forks
- 11
- PR merge metrics
- No merged PRs in 30d
Description
## Summary
Please add a deterministic post-install verification step for NPU readiness.
## Why
Current install/debug flow can be confusing when some diagnostics fail/noise appears, while NPU is actually usable. A single canonical check would reduce support overhead.
## Proposed validation
1. Enumerate devices:
- `ov.Core().available_devices`
- confirm `NPU` present
2. Print `FULL_DEVICE_NAME` for NPU
3. Compile and run a minimal OpenVINO model on NPU (e.g., add-one toy graph)
## Expected output
- Devices include `CPU`, `GPU`, `NPU`
- NPU full name shown (e.g., `Intel(R) AI Boost`)
- Minimal inference on NPU succeeds
## What we used successfully
We validated with:
- `/dev/accel/accel0` present
- OpenVINO device list: `['CPU', 'GPU', 'NPU']`
- NPU full name available
- Minimal NPU inference works
This verification helped separate real installer issues from unrelated script-launch artifacts.
Contributor guide
Research direction
Start by locating the existing post-install verification or debug entry point; the issue names no file or test. Run the OpenVINO device enumeration and minimal NPU compile/inference checks, and consider the work complete when CPU, GPU, and NPU are reported, the NPU full name is shown, and the minimal inference succeeds.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- powershell, python
- Domain
- ai, devops
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 52/100