frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Show HN: PgDog – Scale Postgres without changing the app

https://github.com/pgdogdev/pgdog
185•levkk•9h ago•42 comments

Show HN: Babyshark – Wireshark made easy (terminal UI for PCAPs)

https://github.com/vignesh07/babyshark
48•eigen-vector•4h ago•26 comments

Show HN: Sowbot – open-hardware agricultural robot (ROS2, RTK GPS)

https://sowbot.co.uk/
107•Sabrees•9h ago•37 comments

Show HN: AI Timeline – 171 LLMs from Transformer (2017) to GPT-5.3 (2026)

https://llm-timeline.com/
134•ai_bot•16h ago•49 comments

Show HN: WorldCanvas – R/place, but with a real world map as the canvas

https://worldcanvas.art
2•recuerdame•2h ago•1 comments

Show HN: CIA World Factbook Archive (1990–2025), searchable and exportable

https://cia-factbook-archive.fly.dev/
458•MilkMp•1d ago•96 comments

Show HN: Fostrom, an IoT Cloud Platform built for developers

https://fostrom.io/
11•arjunbajaj•3d ago•6 comments

Show HN: PureBee – A software-defined GPU running Llama 3.2 1B at 3.6 tok/SEC

https://github.com/PureBee/purebee
3•benryanx•3h ago•1 comments

Show HN: Peekl – A modern alternative to Ansible and Puppet

https://peekl.dev
2•redat00•3h ago•0 comments

Show HN: Merkle Casino – Random CT Domains

https://merkle.altayakkus.dev
2•biosboiii•3h ago•0 comments

Show HN: Agent Multiplexer – manage Claude Code via tmux

https://github.com/mixpeek/amux
7•Beefin•9h ago•1 comments

Show HN: Search-sessions – Search all your Claude Code session history in <300ms

https://github.com/sinzin91/search-sessions
4•sinzin91•4h ago•4 comments

Show HN: A geometric analysis of Chopin's Prelude No. 4 using 3D topology

https://github.com/jimishol/cholidean-harmony-structure/blob/main/docs/03-case-study-chopin-prelu...
47•jimishol•3d ago•11 comments

Show HN: 3D Mahjong, Built in CSS

https://voxjong.com
125•rofko•1d ago•58 comments

Show HN: Implementing ping from the Ethernet layer (ARP,IPv4,ICMP in user space)

https://github.com/v420v/ping
7•ibuki256•15h ago•1 comments

Show HN: I vibe-coded a custom WebGPU engine for my MMO

https://github.com/verekia/voidcore
3•jverrecchia•6h ago•1 comments

Show HN: BVisor – An Embedded Bash Sandbox, 2ms Boot, Written in Zig

https://github.com/butter-dot-dev/bVisor
13•edunteman•7h ago•4 comments

Show HN: AgentDbg - local-first debugger for AI agents (timeline, loops, etc.)

https://github.com/AgentDbg/AgentDbg
3•z-a-f•7h ago•2 comments

Show HN: Unlock the best engineering knowledge in papers for your coding agent

https://code.paperlantern.ai
5•kalpitdixit•7h ago•20 comments

Show HN: Local-First Linux MicroVMs for macOS

https://shuru.run
207•harshdoesdev•1d ago•62 comments

Show HN: What I've learned from shipping 25 mobile apps

https://newsletter.masilotti.com/p/what-ive-learned-from-shipping-25
3•joemasilotti•8h ago•0 comments

Show HN: Llama 3.1 70B on a single RTX 3090 via NVMe-to-GPU bypassing the CPU

https://github.com/xaskasdf/ntransformer
387•xaskasdf•2d ago•102 comments

Show HN: Free ecommerce platform for link-in-bio people

https://stoar.page/
2•arajnoha•8h ago•2 comments

Show HN: Rendering 18,000 videos in real-time with Python

https://madebymohammed.com/pysaic
40•mbmproductions•1d ago•5 comments

Show HN: Mato – a Multi-Agent Terminal Office workspace (tmux-like)

https://github.com/mr-kelly/mato
3•chepy•9h ago•0 comments

Show HN: TTSLab – A voice AI agent and TTS lab running in the browser via WebGPU

https://ttslab.dev
5•MbBrainz•9h ago•2 comments

Show HN: EloPhanto – A self-evolving AI agent that builds its own tools

https://github.com/elophanto/EloPhanto
2•elophanto_agent•9h ago•0 comments

Show HN: Self-hosted lightweight file sharing app. (folderhost)

https://github.com/MertJSX/folderhost
8•mertjsx•9h ago•0 comments

Show HN: TLA+ Workbench skill for coding agents (compat. with Vercel skills CLI)

https://github.com/younes-io/agent-skills/tree/main/skills/tlaplus-workbench
41•youio•1d ago•4 comments

Show HN: I built an iOS app to WebRTC into my Mac terminal from the toilet

https://macky.dev
2•Sayuj01•10h ago•2 comments
Open in hackernews

Show HN: PureBee – A software-defined GPU running Llama 3.2 1B at 3.6 tok/SEC

https://github.com/PureBee/purebee
3•benryanx•3h ago
This started as a question about simulation theory: if a GPU is just rules applied to a grid in parallel, do you actually need the silicon?

Turns out, no.

PureBee is a complete GPU defined as a software specification — Memory, Engine, Instruction Set, Runtime. It runs Llama 3.2 1B inference at 3.6 tok/sec on a single CPU core. The model answers questions correctly.

What makes it different from llama.cpp or WebLLM:

The WASM compute kernel is constructed byte-by-byte in JavaScript at runtime. No Emscripten. No Rust. No compiler. No build step. The binary that runs the Q4 SIMD matrix math is itself readable JavaScript. Every layer of the stack — including the thing executing the math — is auditable source.

The progression from first principles:

```

Baseline JS 0.08 tok/sec

Typed arrays 0.21 tok/sec

WASM kernels 0.70 tok/sec

Q4 quantization 1.30 tok/sec

SIMD 3.00 tok/sec

Worker threads 3.60 tok/sec

```

45× total. Single CPU core. Zero npm dependencies.

The claim isn't that this is faster than a real GPU. The claim is that a GPU was never the hardware — it was always the math. The hardware is just one way to run the math fast. PureBee is another way. If that's true, it changes where inference can run.

To run:

```

git clone https://github.com/PureBee/purebee

node download.js llama3

node --max-old-space-size=4096 chat-llama3.js

```

Requires Node.js ≥ 20. The heap flag is not optional.

Licensed FSL-1.1 (converts to Apache 2.0 in 2 years). Free for personal and internal use.

Happy to go deep on the WASM binary construction, the Q4 nibble layout, or the SharedArrayBuffer weight cache that runs a 4.5GB model in 1.8GB of RAM.

Comments

benryanx•3h ago
The part I'd point people to first is ARCHITECTURE.md — specifically the WASM binary construction section. Every other CPU inference project I know of uses Emscripten or a compiled Rust backend. PureBee builds the binary in JavaScript. That's the thing I'd most want challenged if I'm wrong about it being novel.