How to Solve WorkloadsSessionHost Performance Issues: The Link to AI Work and the Solution

 Hello! Recently, I experienced a frustrating slowdown on my computer while testing AI models. For anyone facing a similar problem, I’ve put together a detailed guide on the cause and how I fixed it.

Press enter or click to view image in full size

1. The Mysterious Process: WorkloadsSessionHost

I was in the middle of a project, using AI models like mstyai and Gemini, when my computer started to feel sluggish. When I opened Task Manager, I discovered the root cause: multiple instances of a process called WorkloadsSessionHost were consuming a huge amount of my system's memory.

This process is typically associated with remote desktop or virtual environments. Since I wasn’t using any of those, I was baffled as to why it kept running automatically.

2. Cause Analysis: The AI Work Was the Culprit

To figure out the identity of WorkloadsSessionHost, I analyzed my system logs and services. I found that this process was deeply connected to a service called WSAIFabricSvc. This service is a core component of Windows 11's AI features, designed to utilize system resources, including the NPU (Neural Processing Unit).

It became clear that my AI testing was the trigger. When my AI models called on Windows’s AI APIs, the WSAIFabricSvc service was automatically activated. To handle the workload, this service then spawned multiple WorkloadsSessionHost processes, causing my system's memory usage to skyrocket.

3. The Solution: A Simple Service Setting Change

Once I understood the cause, the fix was simple and effective.

1. Open the Services App: Press the Windows Key + R, type services.msc, and hit Enter.
2. Find the Service: In the list, locate WSAIFabricSvc and double-click it.
3. Change Startup Type: I found that its startup type was set to Automatic. This was the reason it kept running at boot. I changed the setting to Manual and clicked OK.

After making this change, I rebooted my computer. WorkloadsSessionHost no longer appeared in my task list.

4. Conclusion: Balancing AI Work with System Optimization

This experience taught me that Windows’s hidden AI services can consume significant system resources in unexpected ways during AI-related tasks. By setting the WSAIFabricSvc startup type to Manual, you can prevent unnecessary memory usage at startup.

If your computer is mysteriously slowing down due to WorkloadsSessionHost, I highly recommend trying this method to solve the problem and create a more efficient environment for your AI work!