mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2025-01-12 05:17:21 +01:00
71 lines
2.3 KiB
Markdown
71 lines
2.3 KiB
Markdown
## Overview
|
|
|
|
`llama-rpc-server` allows running `ggml` backend on a remote host.
|
|
The RPC backend communicates with one or several instances of `llama-rpc-server` and offloads computations to them.
|
|
This can be used for distributed LLM inference with `llama.cpp` in the following way:
|
|
|
|
```mermaid
|
|
flowchart TD
|
|
rpcb---|TCP|srva
|
|
rpcb---|TCP|srvb
|
|
rpcb-.-|TCP|srvn
|
|
subgraph hostn[Host N]
|
|
srvn[llama-rpc-server]-.-backend3["Backend (CUDA,Metal,etc.)"]
|
|
end
|
|
subgraph hostb[Host B]
|
|
srvb[llama-rpc-server]---backend2["Backend (CUDA,Metal,etc.)"]
|
|
end
|
|
subgraph hosta[Host A]
|
|
srva[llama-rpc-server]---backend["Backend (CUDA,Metal,etc.)"]
|
|
end
|
|
subgraph host[Main Host]
|
|
ggml[llama.cpp]---rpcb[RPC backend]
|
|
end
|
|
style hostn stroke:#66,stroke-width:2px,stroke-dasharray: 5 5
|
|
```
|
|
|
|
Each host can run a different backend, e.g. one with CUDA and another with Metal.
|
|
You can also run multiple `llama-rpc-server` instances on the same host, each with a different backend.
|
|
|
|
## Usage
|
|
|
|
On each host, build the corresponding backend with `cmake` and add `-DLLAMA_RPC=ON` to the build options.
|
|
For example, to build the CUDA backend with RPC support:
|
|
|
|
```bash
|
|
cmake -B build-rpc-cuda -DLLAMA_CUDA=ON -DLLAMA_RPC=ON
|
|
cmake --build build-rpc-cuda --config Release
|
|
```
|
|
|
|
Then, start `llama-rpc-server` with the backend:
|
|
|
|
```bash
|
|
$ bin/llama-rpc-server -p 50052
|
|
create_backend: using CUDA backend
|
|
ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
|
|
ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes
|
|
ggml_cuda_init: found 1 CUDA devices:
|
|
Device 0: NVIDIA T1200 Laptop GPU, compute capability 7.5, VMM: yes
|
|
Starting RPC server on 0.0.0.0:50052
|
|
```
|
|
|
|
When using the CUDA backend, you can specify the device with the `CUDA_VISIBLE_DEVICES` environment variable, e.g.:
|
|
```bash
|
|
$ CUDA_VISIBLE_DEVICES=0 bin/llama-rpc-server -p 50052
|
|
```
|
|
This way you can run multiple `llama-rpc-server` instances on the same host, each with a different CUDA device.
|
|
|
|
|
|
On the main host build `llama.cpp` only with `-DLLAMA_RPC=ON`:
|
|
|
|
```bash
|
|
cmake -B build-rpc -DLLAMA_RPC=ON
|
|
cmake --build build-rpc --config Release -t -j
|
|
```
|
|
|
|
Finally, use the `--rpc` option to specify the host and port of each `llama-rpc-server`:
|
|
|
|
```bash
|
|
$ bin/llama-cli -m ../models/tinyllama-1b/ggml-model-f16.gguf -p "Hello, my name is" --repeat-penalty 1.0 -n 64 --rpc 192.168.88.10:50052,192.168.88.11:50052 -ngl 99
|
|
```
|