llama.cpp

EDEN KV Cache Quantization

AI-written implementation of EDEN quantized KV cache.

Two new quantization types for the key-value cache:

Based on EDEN: Communication-Efficient and Robust Distributed Mean Estimation for Federated Learning (ICML 2022).

Usage

llama-server \
  --cache-type-v eden4 \       # quantize value cache
  --cache-type-k q8_0 \        # K cache recommended as F16 or q8_0
  -fa on \                     # flash attention required for GPU
  ...

For K cache with EDEN:

  --cache-type-k eden3 --cache-type-v eden3

Note: CUDA converts to F16 internally for flash attention.

Backend Support

Note: vec_dot on CUDA uses __dp4a for eden4 (table lookup + int8 dp4a) and direct float dequant for eden3. MMQ and FATTN direct paths are not yet implemented for EDEN types on CUDA (fall back to generic paths). OpenCL, SYCL, and Metal backends lack EDEN type support entirely - CUDA is the primary GPU target.

Comparison with Turboquant

EDEN achieves the same perplexity as Turboquant with ~80% less code:

Both methods target 4-bit KV cache quantization. EDEN uses a simple Lloyd-Max codebook with a closed-form optimal scale, while Turboquant employs a more complex iterative optimization.

---

Quick start

A few options to get llama.cpp installed on your machine:

• Visit https://llama.app and follow the instructions
• Run with Docker - see our Docker documentation
• Download pre-built binaries from the releases page
• Build from source by cloning this repository - check out our build guide

Once installed:

# Download and run a model directly from Hugging Face
llama cli -hf ggml-org/Qwen3.5-0.8B-GGUF

# Launch OpenAI-compatible API server
llama serve -hf ggml-org/Qwen3.5-0.8B-GGUF

Description

The main goal of llama.cpp is to enable LLM (and VLM) inference with minimal setup and state-of-the-art performance on a wide range of hardware - locally and in the cloud.

• Plain C/C++ implementation without any dependencies
• Apple silicon is a first-class citizen - optimized via ARM NEON, Accelerate and Metal frameworks
• AVX, AVX2, AVX512 and AMX support for x86 architectures
• RVV, ZVFH, ZFH, ZICBOP and ZIHINTPAUSE support for RISC-V architectures
• 1.5-bit, 2-bit, 3-bit, 4-bit, 5-bit, 6-bit, and 8-bit integer quantization for faster inference and reduced memory use
• Custom CUDA kernels for running LLMs on NVIDIA GPUs (support for AMD GPUs via HIP and Moore Threads GPUs via MUSA)
• Vulkan and SYCL backend support
• CPU+GPU hybrid inference to partially accelerate models larger than the total VRAM capacity

The llama.cpp project is build on top of the ggml library.

Supported backends

Documentation

Tools

• cli
• completion
• server
• GBNF grammars

Development

• How to build
• Running on Docker
• Build on Android
• Multi-GPU usage
• Performance troubleshooting
• GGML tips & tricks
• XCFramework
• Completions
• Models
• Release process

Contributing

• Contributors can open PRs
• Collaborators will be invited based on contributions
• Maintainers can push to branches in the llama.cpp repo and merge PRs into the master branch
• Any help with managing issues, PRs and projects is very appreciated!
• Read the CONTRIBUTING.md for more information

Acknowledgements

• yhirose/cpp-httplib - Single-header HTTP server, used by llama-server - MIT license
• nothings/stb - Single-header image format decoder, used by multimodal subsystem - Public domain
• nlohmann/json - Single-header JSON library, used by various tools/examples - MIT License
• mackron/miniaudio - Single-header audio format decoder, used by multimodal subsystem - Public domain
• sheredom/subprocess.h - Single-header process launching solution for C and C++ - Public domain