TL;DR
Play games on Amazon Luna for your next adventure, included with Prime
- A rotating selection of games, no download needed
- Play on TV, laptop or phone
- Fast, free delivery for your gear, too
Microsoft and GitHub executive Scott Hanselman has released gputaskman, a portable GPU-monitoring app available through GitHub. It reports GPU use by model and process across Windows, Linux through WSL, and macOS, offering more detail than a headline memory figure in Task Manager.
Microsoft and GitHub executive Scott Hanselman has released gputaskman, an unofficial GPU-monitoring app that breaks down graphics activity by model and process. The tool is aimed at users who want more detail than Windows 11 Task Manager’s summary figures provide, including people running local AI models and those checking which apps are using GPU resources.
According to XDA Developers, gputaskman supports Nvidia and AMD graphics cards on Windows, Nvidia GPUs on Linux and Windows Subsystem for Linux (WSL), and Apple Silicon on macOS. The app is available from its GitHub repository, which offers portable downloads that can be run without a traditional installation.
The app is designed to help identify what sits behind an overall GPU-memory reading. Its display can distinguish activity associated with model weights and a model’s KV cache from use by a browser, desktop compositor or rendering process. Those details may help users investigate high memory use, but the available description does not claim that gputaskman changes GPU performance or reduces resource consumption.
Hanselman described the problem the app addresses: “Task Manager says ‘15 GB used.’ This tries to answer: by what? Model weights? KV cache? Your browser? The desktop compositor? And why is so much GPU memory in system RAM?” The stated aim is to turn a broad usage figure into a more specific view of which workloads and processes are involved.
A clearer view of GPU use with gputaskman
Scott Hanselman’s unofficial portable app breaks GPU activity down by model and process, helping users investigate what sits behind a broad memory total.
What the app can show
It aims to identify workloads behind GPU use, including local AI models and everyday apps.
Model components
Distinguish activity linked to model weights and KV cache.
Other GPU users
Inspect use associated with browsers, rendering, and the desktop compositor.
Reported platform support
Support varies by operating system and graphics hardware.
| Platform | Reported GPU support |
|---|---|
| Windows | Nvidia and AMD |
| Linux and WSL | Nvidia |
| macOS | Apple Silicon |
Getting started
Portable downloads are available from the project’s GitHub repository.
Review the repository for current build details and requirements.
Run a supported download without a traditional installation.
Use model and process details as a starting point for investigating GPU memory use.
More Detail on GPU Memory Use
A single GPU-memory total can be difficult to act on: it tells users that resources are occupied, but not necessarily which program or workload is responsible. Process- and model-level information can give users a clearer starting point when investigating memory pressure, especially if they are running local language models alongside everyday applications.
That use case has become more visible as people run large language models on personal computers, where models and their supporting data can consume substantial memory. A tool that separates model activity from browsers, rendering and desktop components may help users understand what is happening before they close applications or change their setup. The actual benefit will depend on how well the app works with a particular GPU, driver and workload.
For Windows users, the release also offers a third-party alternative for inspecting GPU activity without installing a full desktop suite, since portable builds are offered. Its availability on Linux, WSL and Apple Silicon Macs broadens the potential audience, though support differs by platform and GPU type.
Products Worth Considering
LG UltraGear 32GS60QC-B 32-inch Curved QHD Gaming Monitor (180Hz, 1ms, 1000R, AMD FreeSync, HDR10)
LG UltraGear 34GX90SA-W 34-inch WQHD Curved OLED Gaming Monitor
Kindle Paperwhite 16GB (2024) – 7 Glare-Free Display, Weeks Battery Life – Black
As an affiliate, we earn on qualifying purchases.
Task Manager’s Usage Breakdown
Windows 11 includes GPU information in Task Manager, but the figure that draws attention is often a total amount of memory in use. That summary does not necessarily identify the individual applications or AI components behind it. gputaskman is intended to fill that visibility gap, rather than replace Windows’ built-in monitoring tools.
XDA Developers identifies Hanselman as a current Microsoft and GitHub executive and reports that he created the app. The project is presented as an unofficial utility, not as a Windows feature or an announced Microsoft product. Its GitHub distribution also means users can review the project page and choose a build for their supported platform before trying it.
“Task Manager says ‘15 GB used.’ This tries to answer: by what? Model weights? KV cache? Your browser? The desktop compositor? And why is so much GPU memory in system RAM?”
— Scott Hanselman, as quoted in the app description
Compatibility and Reporting Limits
The available information does not specify which GPU models, driver versions or operating-system releases are supported beyond the broad platform and vendor descriptions. It also does not provide independent testing of the readings, comparisons with other monitoring apps, or measured effects on performance and battery life. Users should check the project’s GitHub page for current build details and any platform-specific requirements.
It is also not clear how completely the app can attribute memory used by every application or AI workload. GPU reporting can vary by hardware and software, so the listed compatibility should not be taken as a guarantee that every process will appear with the same level of detail on every system.
Check the GitHub Builds
People interested in trying gputaskman can visit its GitHub repository and select a portable download for a supported platform. The next useful information for prospective users will be the project’s own documentation and updates, particularly any notes on compatibility, known limitations and changes to the monitoring display.
No broader Microsoft release or formal Windows integration is described. For now, gputaskman remains an unofficial, independently distributed utility; its practical value will depend on how reliably it reports activity on users’ specific systems.
Source: XDA Developers
Key Questions
What is gputaskman?
gputaskman is an unofficial GPU-monitoring app created by Scott Hanselman. It aims to show which models and processes are using GPU resources, rather than presenting only a broad memory total.
Which platforms does it support?
The reported support includes Nvidia and AMD GPUs on Windows, Nvidia GPUs on Linux and WSL, and Apple Silicon on macOS. Check the GitHub repository for build-specific requirements.
Can it identify GPU use from local AI models?
That is one of its stated uses. The app is intended to show activity associated with individual models and components such as model weights and KV cache, along with other processes.
Does gputaskman need to be installed?
Portable downloads are available through the GitHub repository, so a traditional installation is not required for those builds.
Is gputaskman an official Windows feature?
No. It is described as an unofficial utility created by a Microsoft and GitHub executive, not as a feature announced by Microsoft or built into Windows 11.
Source: XDA Developers
