hsfzxjy/npunlock
Write and run custom C kernels for Intel Core Ultra NPUs
About the project
A toolkit for writing and running custom C kernels on Intel Core Ultra NPUs, bypassing the public graph-level-only stack. It reconstructs the path from C code to a runnable NPU kernel, verified on Windows x64 with Meteor Lake / NPU3720.
Useful for
- Run a custom FP32 GELU kernel on NPU3720 and verify against NumPy
- Embed a C kernel into an NPU graph via npunlock Python bindings
- Call libm functions such as tanhf inside a custom NPU kernel
README summarized by DeepSeek V4.1 Flash. Details may be inaccurate.
Why it’s trending
- 16 stars so far today, about 31 expected by the end of the day.
- The spike has held for 2 days in a row — not a one-off blip.
- The repository is 3 days old and already has 36 stars. With less than two weeks of history, there's no usual pace to compare the spike against yet.
- Hacker News: “Show HN: Npunlock – Run custom C kernels for Intel NPUs” — 24 points, 21 h ago.
Stars per day
Bars are daily stars, the line is the usual pace. Red marks spike days.
Numbers
- Total stars
- 36
- Today
- 16 · ≈ 31 by evening
- Forks
- 0
- Issues and pull requests
- 0
- Watchers
- 1
- Language
- C
- License
- Apache-2.0
- Created
- September 20, 2026
- Last push
- September 23, 2026
Star trust
Growth looks organic: forks and discussion are in line with active projects, and stars arrive unevenly, the way people give them.
These are heuristics, not a verdict: we judge by the repository’s behavior, not by a list of stargazers.
Hacker News discussions
- Show HN: Npunlock – Run custom C kernels for Intel NPUs24 points, September 22, 2026
Spotted in
- September 23, 2026Spotted on Hacker News
More in this category
-
5.3
xCuri0/ReBarUEFI
UEFI DXE driver that enables Resizable BAR on systems without official support. Provides GPU performance gains and is required for Intel Arc GPUs to work optimally.
-
4.6
General-Instinct/InstinctFlash
A high-performance serving runtime for robotics models (VLA, WAM, policies) with FP8 acceleration and optimized kernels on Jetson Thor and RTX 4090/5090.
-
2.9
lightorigins/Light-O1
Light-O1-Preview is a text-to-action model that turns natural-language instructions into whole-body humanoid motion, outputting a (frames, 138) action array at 20 FPS. It is pretrained on human action recovered from…
-
2.8
Lakr233/vphone-cli
CLI tool for booting a virtual iPhone on Apple Silicon via Virtualization.framework using PCC research VM infrastructure, including firmware patching and CFW installation.
-
2.7
ruvnet/RuView
A Rust WiFi-sensing platform that uses CSI data from cheap ESP32 boards to detect presence, breathing, heart rate, activity and pose without cameras or wearables. Runs locally on edge hardware with Home Assistant,…
-
2.6
amagine-ai/Amagine3D
An open-source 3D design layer for hardware creation: given a product description, reference images and dimensions, it generates editable enclosures and assemblies with STEP, STL and 3MF export.