vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update README.md

Browse source
This commit is contained in:
calvinweb 2024-02-03 17:06:40 +08:00 • committed by GitHub
parent b51e79422c
commit 64e2bc9184
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 6 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

8
README.md
View file

@ -346,11 +346,15 @@ mpirun -hostfile hostfile -n 3 ./main -m ./models/7B/ggml-model-q4_0.gguf -n 128
``` ```
### Vulkan Build ### Vulkan Build
**Vulkan** is a new library aimed at replacing **OpenGL**. Starting from version `b1996`, `llama.cpp` now supports **Vulkan** as a backend. It can be faster (almost 2x on NVIDIA GeForce RTX™ 3060 Laptop GPU) than **CUDA** on a consumer-grade NVIDIA graphics card. **Vulkan** is a new library aimed at replacing **OpenGL**. Starting from version `b1996`, `llama.cpp` now supports **Vulkan** as a backend.
To build with **Vulkan**: To build with **Vulkan**:
- Ensure that your operating system and graphics card support **Vulkan**. - Ensure that your operating system and graphics card support **Vulkan**.
- Install Vulkan realted dependents:
- On Debian based distros:
```
apt install libvulkan-dev
```
- Build `llama.cpp` with the following arguments: `LLAMA_VULKAN=1`. - Build `llama.cpp` with the following arguments: `LLAMA_VULKAN=1`.
### BLAS Build ### BLAS Build