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

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
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
12 changed files with 1395 additions and 98 deletions
Download patch file Download diff file Expand all files Collapse all files

10
common/common.cpp
View file

@ -1060,6 +1060,14 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
#endif // GGML_USE_CUDA_SYCL_VULKAN
return true;
}
if (arg == "--rpc") {
if (++i >= argc) {
invalid_param = true;
return true;
}
params.rpc_servers = argv[i];
return true;
}
if (arg == "--no-mmap") {
params.use_mmap = false;
return true;
@ -1557,6 +1565,7 @@ void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
printf(" -mg i, --main-gpu i the GPU to use for the model (with split-mode = none),\n");
printf(" or for intermediate results and KV (with split-mode = row) (default: %d)\n", params.main_gpu);
}
printf(" --rpc SERVERS comma separated list of RPC servers\n");
printf(" --verbose-prompt print a verbose prompt before generation (default: %s)\n", params.verbose_prompt ? "true" : "false");
printf(" --no-display-prompt don't print prompt at generation (default: %s)\n", !params.display_prompt ? "true" : "false");
printf(" -gan N, --grp-attn-n N\n");
@ -1830,6 +1839,7 @@ struct llama_model_params llama_model_params_from_gpt_params(const gpt_params &
if (params.n_gpu_layers != -1) {
mparams.n_gpu_layers = params.n_gpu_layers;
}
mparams.rpc_servers = params.rpc_servers.c_str();
mparams.main_gpu = params.main_gpu;
mparams.split_mode = params.split_mode;
mparams.tensor_split = params.tensor_split;