Llm.sql – Run a 640MB LLM on SQLite, with 210MB peak RSS and 7.4 tok/s
Hi HN, I built llm.sql, an LLM inference framework that reimagines the LLM execution pipeline as a series of structured SQL queries atop SQLite. The motivation: Edge LLMs are getting better, but hardware remains a bottleneck, especially RAM (size and bandwidth). When available memory is less than the model size and KV cache, the OS incurs page faults and swaps pages using LRU-like strategies, resulting in throughput degradation that's hard to notice and even harder to debug. In fact, the memory access pattern during LLM inference is deterministic - we know exactly which weights are needed…
What it does
In the maker’s words, at launch
Hi HN, I built llm.sql, an LLM inference framework that reimagines the LLM execution pipeline as a series of structured SQL queries atop SQLite. The motivation: Edge LLMs are getting better, but hardware remains a bottleneck, especially RAM (size and bandwidth). When available memory is less than the model size and KV cache, the OS incurs page faults and swaps pages using LRU-like strategies, resulting in throughput degradation that's hard to notice and even harder to debug. In fact, the memory access pattern during LLM inference is deterministic - we know exactly which weights are needed and when. This means even Bélády's optimal page replacement algorithm is applicable here. So instead of letting the OS manage memory, llm.sql takes over: - Model parameters are stored in SQLite BLOB tables - Computational logic is implemented as SQLite C extensions - Memory management is handled explicitly, not by the OS - Zero heavy dependencies. No PyTorch, no Transformers. Just Python, C, or C++ This gives us explicit, deterministic control over what's in memory at each step of inference. Results: Running Qwen2.5-0.5B-INT8 (~640MB model) with a peak RSS ~210MB and 7.40 tokens/s throughput. Alpha version is available on GitHub: https://github.com/xuxianghong12/llm.sql I'm the developer, happy to answer any technical questions about the design and implementation.
Does the same job
all alternatives →- SUSpeeding up LLM inference 2x times (possibly)2024 · asciinema.org · ▲419
Here's a project I've been working on for the last few months. It's a new (I think) algorithm, that allows to adjust smoothly - and in real time - how many calculations you'd like to do during inference of an LLM model. It seems that it's possible to do just 20-25% of weight multiplications instead of all of them, and still get good inference results. I implemented it to run on M1/M2/M3 GPU. The mmul approximation itself can be pushed to run 2x fast before the quality of output collapses. The inference speed is just a bit faster than Llama.cpp's, because the rest of implementation…
- ALA local-first memory store for LLM agents (SQLite)Dec 2025 · github.com · ▲48
- PTPredictive text using only 13kb of JavaScript. no LLM2024 · adamgrant.info · ▲207
- LALLM, a Rust Crate/CLI for CPU Inference of LLMs (LLaMA, GPT-NeoX, etc.)2023 · github.com · ▲45
G'day, HN! I'm one of the maintainers of `llm`. I've been working alongside a trusty group of contributors to bring this project to life, and we're now at a point where we're ready to share it with the world. Large language models (LLMs) are taking the computing world by storm due to their emergent abilities that allow them to perform a wide variety of tasks, including translation, summarization, code generation, and even some degree of reasoning. However, the ecosystem around LLMs is still in its infancy, and it can be difficult to get started with these models. `llm` is a one-stop shop for…
- TVTiny-vLLM – high performance LLM inference engine in C++ and CUDAMay 2026 · github.com · ▲205
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 27d ago · cactuscompute.com
Launched alongside, April 2026
the whole month →
- AG
Thought the resources for GPU arch were lacking, so here we are
Life & fun · Apr 2026 · jaso1024.com
- IB
Built a ~9M param LLM from scratch to understand how they actually work. Vanilla transformer, 60K synthetic conversations, ~130 lines of PyTorch. Trains in 5 min on a free Colab T4. The fish thinks the meaning of life is food. Fork it and swap the personality for your own character.
AI · Apr 2026 · github.com
- BC
Life & fun · Apr 2026 · sam-burns.com
- IB
With social media and now AI, its important to keep the indie web alive. There are many people who write frequently. Blogosphere tries to highlight them by fetching the recent posts from personal blogs across many categories. There are two versions: Minimal (HN-inspired, fast, static): https://text.blogosphere.app/ Non-minimal: https://blogosphere.app/ If you don't find your blog (or your favorite ones), please add them. I will review and approve it.
AI · Apr 2026 · text.blogosphere.app