Microsoft engineer Scott Hanselman has developed a free Windows utility called gputaskman that provides detailed information about GPU usage, video memory consumption and local AI workloads.
The application is designed to offer more insight than Windows 11's built in Task Manager, particularly when identifying which processes are consuming graphics memory.
While Task Manager can display GPU activity and memory usage for individual applications, it does not always explain how those applications are using the available resources.
Gputaskman addresses this limitation by showing additional information about GPU processes, including memory allocated to AI models, web browsers and desktop rendering.
The utility is currently in early development and supports Windows PCs equipped with AMD or Nvidia graphics processors.
Gputaskman Provides More Detailed GPU and VRAM Monitoring
One of the application's main features is its ability to explain how graphics memory is being used by different workloads.
For example, Windows Task Manager might report that 15GB of GPU memory is occupied without clearly identifying how much belongs to an AI model or other processes.
Gputaskman attempts to provide a more useful breakdown of that consumption.
The utility can identify local large language models and display information about their memory requirements and related GPU resources.
It also helps distinguish between memory used for model weights, key value caches, browser activity and desktop composition.
Another feature is GPU temperature monitoring, allowing you to observe thermal conditions alongside memory consumption.
These details could be particularly useful when troubleshooting performance issues or determining why a graphics card has less available memory than expected.
Windows GPU Monitoring Tools Compared
| Feature | Windows Task Manager | Gputaskman |
|---|---|---|
| GPU usage by process | Yes | Yes |
| VRAM consumption | Yes | Yes |
| Local AI model identification | Limited | Yes |
| AI memory breakdown | Limited | Yes |
| GPU temperature | Not typically displayed | Yes |
| AMD and Nvidia GPU support | Yes | Yes |
Gputaskman is available through GitHub, where Hanselman has invited developers to contribute improvements through pull requests.
Although the application currently supports AMD and Nvidia graphics hardware, broader compatibility remains a development goal.
Hanselman has specifically mentioned plans to support devices such as the Surface Laptop Ultra, RTX Spark systems and other Windows on Arm computers.
Those capabilities are not yet confirmed as available in the current version.
Local AI Workloads Make Detailed GPU Monitoring More Useful
The utility arrives as more Windows applications begin running AI workloads directly on personal computers.
Local language models can consume substantial amounts of GPU memory, particularly when loading model weights and maintaining conversation context.
When dedicated VRAM becomes insufficient, some workloads may also use shared system memory, potentially affecting performance.
Microsoft's recently announced Hybrid intelligence initiative combines local AI processing with cloud services, increasing the importance of understanding hardware resource allocation.
Gputaskman provides an additional diagnostic option for developers, AI enthusiasts and PC owners who want more detailed information about graphics memory usage.
However, because it remains experimental, compatibility and available features may change as development continues.
Discussion (0)
Be the first to comment.