CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-03 09:22:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml : add RPC backend (#6829)

Browse Source 
* ggml : add RPC backend

The RPC backend proxies all operations to a remote server which runs a
regular backend (CPU, CUDA, Metal, etc).

* set TCP_NODELAY

* add CI workflows

* Address review comments

* fix warning

* implement llama_max_devices() for RPC

* Address review comments

* Address review comments

* wrap sockfd into a struct

* implement get_alignment and get_max_size

* add get_device_memory

* fix warning

* win32 support

* add README

* readme : trim trailing whitespace

* Address review comments

* win32 fix

* Address review comments

* fix compile warnings on macos
This commit is contained in:
Radoslav Gerganov
2024-05-14 14:27:19 +03:00
committed by GitHub
parent 541600201e
commit 5e31828d3e
12 changed files with 1395 additions and 98 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
examples/CMakeLists.txt
View File

@@ -49,4 +49,7 @@ else()
add_subdirectory(server)
endif()
add_subdirectory(export-lora)
if (LLAMA_RPC)
add_subdirectory(rpc)
endif()
endif()

2
examples/rpc/CMakeLists.txt Normal file
View File

@@ -0,0 +1,2 @@
add_executable(rpc-server rpc-server.cpp)
target_link_libraries(rpc-server PRIVATE ggml llama)

74
examples/rpc/README.md Normal file
View File

@@ -0,0 +1,74 @@
## Overview
The `rpc-server` allows running `ggml` backend on a remote host.
The RPC backend communicates with one or several instances of `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[rpc-server]-.-backend3["Backend (CUDA,Metal,etc.)"]
end
subgraph hostb[Host B]
srvb[rpc-server]---backend2["Backend (CUDA,Metal,etc.)"]
end
subgraph hosta[Host A]
srva[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 `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
mkdir build-rpc-cuda
cd build-rpc-cuda
cmake .. -DLLAMA_CUDA=ON -DLLAMA_RPC=ON
cmake --build . --config Release
```
Then, start the `rpc-server` with the backend:
```bash
$ bin/rpc-server 0.0.0.0 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/rpc-server 0.0.0.0 50052
```
This way you can run multiple `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
mkdir build-rpc
cd build-rpc
cmake .. -DLLAMA_RPC=ON
cmake --build . --config Release
```
Finally, use the `--rpc` option to specify the host and port of each `rpc-server`:
```bash
$ bin/main -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
```

70
examples/rpc/rpc-server.cpp Normal file
View File

@@ -0,0 +1,70 @@
#ifdef GGML_USE_CUDA
#include "ggml-cuda.h"
#endif
#ifdef GGML_USE_METAL
#include "ggml-metal.h"
#endif
#include "ggml-rpc.h"
#include <string>
#include <stdio.h>
static ggml_backend_t create_backend() {
ggml_backend_t backend = NULL;
#ifdef GGML_USE_CUDA
fprintf(stderr, "%s: using CUDA backend\n", __func__);
backend = ggml_backend_cuda_init(0); // init device 0
if (!backend) {
fprintf(stderr, "%s: ggml_backend_cuda_init() failed\n", __func__);
}
#elif GGML_USE_METAL
fprintf(stderr, "%s: using Metal backend\n", __func__);
backend = ggml_backend_metal_init();
if (!backend) {
fprintf(stderr, "%s: ggml_backend_metal_init() failed\n", __func__);
}
#endif
// if there aren't GPU Backends fallback to CPU backend
if (!backend) {
fprintf(stderr, "%s: using CPU backend\n", __func__);
backend = ggml_backend_cpu_init();
}
return backend;
}
static void get_backend_memory(size_t * free_mem, size_t * total_mem) {
#ifdef GGML_USE_CUDA
ggml_backend_cuda_get_device_memory(0, free_mem, total_mem);
#else
// TODO: implement for other backends
*free_mem = 1;
*total_mem = 1;
#endif
}
int main(int argc, char * argv[]) {
if (argc < 3) {
fprintf(stderr, "Usage: %s <host> <port>\n", argv[0]);
return 1;
}
const char * host = argv[1];
int port = std::stoi(argv[2]);
if (port <= 0 || port > 65535) {
fprintf(stderr, "Invalid port number: %d\n", port);
return 1;
}
ggml_backend_t backend = create_backend();
if (!backend) {
fprintf(stderr, "Failed to create backend\n");
return 1;
}
printf("Starting RPC server on %s:%d\n", host, port);
size_t free_mem, total_mem;
get_backend_memory(&free_mem, &total_mem);
std::string endpoint = std::string(host) + ":" + std::to_string(port);
start_rpc_server(backend, endpoint.c_str(), free_mem, total_mem);
ggml_backend_free(backend);
return 0;
}