# FeSens/openTPU

> An open-source AI accelerator on FPGA: RTL, ISA, simulator, compiler and profiler in one repo, running Qwen3, LFM2.5 and other models on a Kintex-7 card.

- Magnitude: 6.8 out of 10 — Early signal
- Stars: 208 total · Not enough measurements today
- Star trust: star growth looks organic
- Category: Hardware, IoT & robotics · Language: Verilog · License: Apache-2.0 · Created: 2026-09-24 · Last push: 2026-10-02
- GitHub: https://github.com/FeSens/openTPU · Page: https://gitnova.dev/en/r/FeSens/openTPU

## Useful for

- Study how an AI accelerator works from Python matmul down to RTL
- Run inference of small LLMs on a Kintex-7 FPGA card
- Compare model performance between the simulator and the real card

## Why it’s here

- There are not enough fresh counter measurements for today's UTC day. The gain is unknown.
- The repository is 13 days old and already has 208 stars. With less than two weeks of history, there's no usual pace to compare the spike against yet.
- Hacker News: “AI is now capable of developing its own inference hardware” — 212 points, 8 h ago.
- Recent forks include notable developers: @mbrukman (394 followers).

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 8
- Issues and pull requests: 0
- Watchers: 0
- Average over the last week: 51 per day
- Usual pace: too little history (under two weeks)
- Stars in the last hour (measured): 13

## Stars per day, last 18 days (oldest → newest, today is partial)

2026-09-20 … 2026-10-07: 0, 0, 0, 0, 1, 0, 1, 10, 10, 5, 2, 0, 2, 1, 0, 0, 181, None

## Magnitude by day, last 2 days

10-06 5.1, 10-07 6.8

## Hacker News

- AI is now capable of developing its own inference hardware — 212 points, 279 comments: https://news.ycombinator.com/item?id=49980715

## Spotted in now

- Spotted on Hacker News

## More in this category

1. **zer0days-op/OpenGoldHEN** — 5.5 · Early signal · Hardware, IoT & robotics · C · Not enough measurements today
   Open-source release of the GoldHEN codebase, a homebrew enabler (HEN) for PlayStation 4 consoles written in C. The README is a manifesto and does not describe features, build steps, or usage.
   Full card: https://gitnova.dev/en/r/zer0days-op/OpenGoldHEN.md
2. **facebookincubator/muse-gadget-sdk** — 4.7 · Cooling · Hardware, IoT & robotics · C · Not enough measurements today
   SDK and firmware for building your own Muse gadgets on ESP32 or Linux: displays, buttons, sensors and audio that pair with the Muse app on iOS and Android.
   Full card: https://gitnova.dev/en/r/facebookincubator/muse-gadget-sdk.md
3. **isod89/sloop-fm1** — 4.4 · Early signal · Hardware, IoT & robotics · C · Not enough measurements today
   A firmware for the M-VAVE FM-1 that turns it into a four-track groovebox: three synths and a drum machine with 16 sounds, nine synthesis engines, user samples and a live song mode. A GPL-3.0 fork of Felucca.
   Full card: https://gitnova.dev/en/r/isod89/sloop-fm1.md
4. **KingKongRobotics/jumper** — 3.6 · Early signal · Hardware, IoT & robotics · Python · Not enough measurements today
   Open-source 22-DoF crab robot Jumper: this repository covers motion training, replay and policy export, plus links to tools for designing appearances and scenes.
   Full card: https://gitnova.dev/en/r/KingKongRobotics/jumper.md
5. **hugelton/Felucca** — 3.4 · Early signal · Hardware, IoT & robotics · C · Not enough measurements today
   Custom firmware for the M-VAVE FM-1 synthesizer: nine engines, four tracks, a 64-step sequencer, effects and a web editor. Installed over USB from a browser with no extra hardware.
   Full card: https://gitnova.dev/en/r/hugelton/Felucca.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-07 00:30 UTC, updated every 30 minutes.