Lightweight Llama3 Inference Engine – CUDA C
Hey, recently I took inspiration from llama.cpp, ollama, and many other similar tools that enable inference of LLMs locally, and I just finished building a Llama inference engine for the 8B model in CUDA C. I recently wanted to explore my newly founded interest in CUDA programming and my passion for machine learning. This project only makes use of the native CUDA runtime api and cuda_fp16. The inference takes place in fp16, so it requires around 17-18GB of VRAM (~16GB for model params and some more for intermediary caches). It doesn’t use cuBLAS or any similar libraries since I wanted to be…
Lightweight Llama3 Inference Engine – CUDA C launched on January 4, 2025 with 12 votes, #1,233 of 6,142 launches that month and more than 75% of that year's launches. The site was still online in September 2026, 1 year on.
In plain words
Lightweight Llama3 Inference Engine is a CUDA C implementation for running the Llama 8B model locally on GPUs with at least 17-18GB of VRAM. It uses only the native CUDA runtime API and half-precision floating point, without relying on optimized libraries like cuBLAS. Designed for developers interested in low-level GPU programming and machine learning, it reads model files from HuggingFace in safetensor format. While less optimized than existing inference engines, it prioritizes educational value and direct exposure to CUDA fundamentals.
written from the facts on this page · September 2026
From the sources
Lightweight Llama 3 8B Inference Engine in CUDA C. Contribute to abhisheknair10/llama3.cu development by creating an account on GitHub.
Llama3.cu is a CUDA native implementation of the LLaMA3 architecture for causal language modeling. Core principles of the transformer architecture from the papers Attention is All You Need and LLaMA: Open and Efficient Foundation Language Models are implemented using custom CUDA kernel definitions, enabling scalable parallel processing on Nvidia GPUs. The models are expected to be downloaded off of HuggingFace. They are stored as BF16 parameter weights in a .safetensor file, which during load time to the CUDA device, is converted to FP16 via a FP32 proxy. Hence, a CUDA device with a minimum of 24GB VRAM must be used. - 24GB+ VRAM CUDA Device - HuggingFace account - Operating System: UNIX or…from github.com
In the maker’s words, at launch
Hey, recently I took inspiration from llama.cpp, ollama, and many other similar tools that enable inference of LLMs locally, and I just finished building a Llama inference engine for the 8B model in CUDA C. I recently wanted to explore my newly founded interest in CUDA programming and my passion for machine learning.…
Does a similar job
all alternatives →
Tiny-vLLM – high performance LLM inference engine in C++ and CUDAMay 2026 · github.com · ▲205Build your own high performance LLM inference engine in C++ and CUDA - a smaller version of vLLM - jmaczan/tiny-vllm
Finetune LLaMA-7B on commodity GPUs using your own text2023 · github.com · ▲449I've been playing around with https://github.com/zphang/minimal-llama/ and https://github.com/tloen/alpaca-lora/blob/main/finetune.py, and wanted to create a simple UI where you can just paste text, tweak the parameters, and finetune the model quickly using a modern GPU. To prepare the data, simply separate your text with two blank lines. There's an inference tab, so you can test how the tuned model behaves. This is my first foray into the world of LLM finetuning, Python, Torch, Transformers, LoRA, PEFT, and Gradio. Enjoy!
Speeding up LLM inference 2x times (possibly)2024 · asciinema.org · ▲419Here'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…
80% faster, 50% less memory, 0% loss of accuracy Llama finetuning2023 · github.com · ▲385Hi HN! I'm just sharing a project I've been working on during the LLM Efficiency Challenge - you can now finetune Llama with QLoRA 5x faster than Huggingface's original implementation on your own local GPU. Some highlights: 1. Manual autograd engine - hand derived backprop steps. 2. QLoRA / LoRA 80% faster, 50% less memory. 3. All kernels written in OpenAI's Triton language. 4. 0% loss in accuracy - no approximation methods - all exact. 5. No change of hardware necessary. Supports NVIDIA GPUs since 2018+. CUDA 7.5+. 6. Flash Attention support via Xformers. 7. Supports 4bit and 16bit…
We made our own inference engine for Apple Silicon2025 · github.com · ▲186We wrote our inference engine on Rust, it is faster than llama cpp in all of the use cases. Your feedback is very welcomed. Written from scratch with idea that you can add support of any kernel and platform.
More ai this month
the category →
Paintings by AI models: programs run through a simulation of oil paint.
AI · 20d ago · stillwet.art
PixVerse R2▲381A real-time world model you can explore and change
AI · 29d ago · world.pixverse.video
Source: Show HN post ↗ · launch text quoted in part
Launched alongside, January 2025
the whole month →
Hello! I'm Byran. I spent the past ~6 months engineering a laptop from scratch. It's fully open-source on GH at: https://github.com/Hello9999901/laptop
Dev tools · 2025 · byran.ee
- TITetris in a PDF▲1,289
I realized that the PDF engines of modern desktop browsers (PDFium and PDF.js) support JavaScript with enough I/O primitives to make a basic game like Tetris. It was a bit tricky to find a union of features that work in both engines, but in the end it turns out that showing/hiding annotation "fields" works well to make monochrome pixels, and keyboard input can be achieved by typing in a text input box. All in all it's quite janky but a nice reminder of how general purpose PDF scripting can be. The linked PDF is all ASCII so you can just open it in a text editor, or have a look at…
Life & fun · 2025 · th0mas.nl
21st.dev▲1,211I originally built this app to organize and reuse interesting React + Tailwind components I came across on Twitter, inspired by the shadcn/ui approach. Later, a friend started contributing their own components, and it got me thinking—this could grow into something bigger, like an NPM for design engineers to share and use components. Would love to hear your thoughts!
Dev tools · 2025 · 21st.dev
Create lifelike, personalized AI avatars from text prompts
AI · 2025 · jogg.ai
CSS library based on Counter Strike 1.6 UI.
Life & fun · 2025 · cs16.samke.me