Skip to content

Foundry Local 0.10.3 does not detect AMD Ryzen AI NPU or register VitisAIExecutionProvider on HP OMEN 16 #1122

Description

@DanaS4F

Problem

I am using an HP OMEN 16-xf0017ax with an AMD Ryzen 9 7940HS, AMD Radeon 780M iGPU, NVIDIA GeForce RTX 4070 Laptop GPU, and an AMD Ryzen AI NPU.

I installed Foundry Local 0.10.3 for Windows x64 (WinML). Foundry Local detects and starts normally, but it does not show any NPU models.

Running:

foundry model ls --device npu

returns an empty table with no NPU models.

I also checked the Foundry cache. The server downloads/registers these execution providers:

  • CUDAExecutionProvider — 100%
  • WebGpuExecutionProvider — 100%
  • NvTensorRTRTXExecutionProvider — 100%

However, VitisAIExecutionProvider is not downloaded or registered.

I also checked the installed Windows AppX packages. The only WinML execution-provider packages installed are NVIDIA TensorRT-RTX packages. There is no AMD/VitisAI WinML execution-provider package installed.

Get-AppxPackage *AMD.NPU* returns nothing.

Get-AppxPackage *WinML* only shows:

  • MicrosoftCorporationII.WinML.NVIDIA.TRT-RTX.EP.1.8 — 1.8.24.0
  • MicrosoftCorporationII.WinML.NVIDIA.TRT-RTX.EP.2 — 2.30.43.0

My AMD software version is 25.9.1.

The AMD NPU is present in the system, but Foundry Local does not expose any NPU model and does not appear to download/register VitisAIExecutionProvider.

This seems very similar to issue #995, but I am using an AMD Ryzen 9 7940HS / OMEN 16 rather than the HP EliteBook mentioned there.

Steps to reproduce

  1. Install Foundry Local 0.10.3 WinML for Windows x64.
  2. Start Foundry Local.
  3. Run:
    foundry cache ls
  4. Run:
    foundry model ls --device npu

Expected behavior

Foundry Local should detect the AMD NPU and make compatible NPU model variants available through the WinML/VitisAI execution provider.

Actual behavior

No NPU models are listed, and VitisAIExecutionProvider is not downloaded or registered.

System information

  • Laptop: HP OMEN 16-xf0017ax
  • CPU: AMD Ryzen 9 7940HS
  • iGPU: AMD Radeon 780M
  • dGPU: NVIDIA GeForce RTX 4070 Laptop GPU
  • NPU: AMD Ryzen AI NPU
  • OS: Windows 11 Pro 25H2
  • Architecture: x64
  • Foundry Local: 0.10.3
  • Backend: WinML
  • AMD Software: 25.9.1
  • Installation: Released package/binary

Could you please confirm whether AMD Ryzen 9 7940HS / Ryzen AI NPU is currently supported by Foundry Local 0.10.3 on Windows x64, and if so, why VitisAIExecutionProvider is not being downloaded/registered on this system?

Diagnostics

System

  • OS: Microsoft Windows 10.0.26200
  • Architecture: X64
  • CPU: AMD Ryzen 9 7940HS w/ Radeon 780M Graphics (16 logical cores)
  • GPU: AMD AMD Radeon(TM) Graphics (512.0 MB), NVIDIA NVIDIA GeForce RTX 4070 Laptop GPU (7.8 GB)
  • NPU: AMD NPU Compute Accelerator Device
  • RAM: 7.2 GB available / 15.3 GB total
  • Disk: 24.1 GB free / 145.7 GB total (C:)
  • .NET: .NET 9.0.18

Service

  • State: Ready
  • Web URLs: http://127.0.0.1:59346
  • CLI version: 0.10.3
  • Foundry Local Core: 1.0.0
  • ORT: 1.26.0
  • ORT GenAI: 0.0.0

Models

  • Available variants: 148
  • Loaded: 0
  • Cached: 0

Connectivity

  • Local service: Reachable
  • Model registry: Reachable

Warnings

  • No warnings.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions