Microsoft's Open-Source GPU Taskman Expands Monitoring
Microsoft VP Scott Hanselman released GPU Taskman, an open-source tool offering detailed per-process VRAM monitoring and AI model insights for Windows, Linux, and macOS.

Microsoft Vice President Scott Hanselman has introduced an open-source GPU monitoring utility called GPU Taskman. The tool provides detailed per-process VRAM monitoring and AI model memory insights, offering deeper visibility into GPU resource usage than standard system tools.
Scott Hanselman, a Vice President and Member of Technical Staff at Microsoft and GitHub, revealed the project on GitHub. The initial version 0.3.0 was released under an MIT license. Hanselman demonstrated the tool on his own system, showing that the Windows desktop manager was the most VRAM-intensive application on his RTX 4080 Super. He indicated that such expanded GPU monitoring capabilities could eventually become a standard part of Windows. The application loads in a terminal window and includes a specialized 'Model inspector' for AI workloads.
Technical Features for AI and Gaming Workloads
GPU Taskman delivers granular telemetry for both gaming and emerging AI applications. It displays a comprehensive set of GPU hardware metrics and distinguishes between different types of workloads.
The tool monitors a wide array of GPU statistics. It combines operating system counters with process metadata and runtime logs to provide specific insights, particularly for AI models.
For AI applications, GPU Taskman can identify models and provide details on their architecture, quantization, parameter counts, and estimated memory requirements for components like the KV-cache. It supports popular AI runtimes including llama.cpp, Ollama, and LM Studio. The user interface is divided into three sections: a top area for GPU hardware monitors, a middle list of processes with their runtime and VRAM consumption, and a bottom pane with detailed information for a selected process.
This focus makes it beneficial for power users running multiple GPU-intensive applications, especially AI models. The tool helps diagnose performance issues; excessive VRAM pressure can cause stuttering as applications are forced to use slower system memory. However, GPU Taskman is experimental software. Some monitoring capabilities require further validation across different hardware configurations. It cannot identify individual game textures or pinpoint exactly which resources have been evicted from VRAM. The app notes that GPU-accessible system RAM usage does not automatically indicate a VRAM overflow condition.
Cross-Platform Availability and Distribution
GPU Taskman is designed for broad compatibility across operating systems and hardware. This cross-platform support is a key feature of the open-source project.
The utility supports NVIDIA and AMD GPUs on Windows 10 and 11, Linux, Windows Subsystem for Linux (WSL), and Apple Silicon macOS systems. A recent update, version 0.3.5 released on November 10, added Intel GPU support on Linux. That update also introduced CPU and system RAM monitoring, along with options to record GPU statistics. The tool is distributed under the permissive MIT license. It is available for download via GitHub Releases and the Python Package Index (PyPI). For Windows users, portable builds are offered that do not require a separate Python installation.
Scott Hanselman stated his motivation clearly: "I wanted better visibility into what was consuming my GPU memory, inviting other developers to contribute to the project." The computing industry is increasingly steering data processing toward the GPU. While the Windows Task Manager can technically view GPU processes, it lacks the detailed per-process and AI-specific insights provided by GPU Taskman. The release of this advanced open-source tool could incentivize Microsoft to integrate a more comprehensive GPU monitoring system directly into its operating system's native Task Manager.





