Skip to content

Commit

Permalink
llama-bench : add support for the RPC backend (ggerganov#7435)
Browse files Browse the repository at this point in the history
  • Loading branch information
@rgerganov
rgerganov authored May 29, 2024
1 parent 87bdf2a commit 210d991
Show file tree
Hide file tree
Showing 3 changed files with 35 additions and 2 deletions.
28 changes: 26 additions & 2 deletions examples/llama-bench/llama-bench.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -178,6 +178,7 @@ struct cmd_params {
std::vector<ggml_type> type_v;
std::vector<int> n_threads;
std::vector<int> n_gpu_layers;
std::vector<std::string> rpc_servers;
std::vector<llama_split_mode> split_mode;
std::vector<int> main_gpu;
std::vector<bool> no_kv_offload;
Expand All @@ -202,6 +203,7 @@ static const cmd_params cmd_params_defaults = {
/* type_v */ {GGML_TYPE_F16},
/* n_threads */ {cpu_get_num_math()},
/* n_gpu_layers */ {99},
/* rpc_servers */ {""},
/* split_mode */ {LLAMA_SPLIT_MODE_LAYER},
/* main_gpu */ {0},
/* no_kv_offload */ {false},
Expand Down Expand Up @@ -230,6 +232,7 @@ static void print_usage(int /* argc */, char ** argv) {
printf(" -ctv, --cache-type-v <t> (default: %s)\n", join(transform_to_str(cmd_params_defaults.type_v, ggml_type_name), ",").c_str());
printf(" -t, --threads <n> (default: %s)\n", join(cmd_params_defaults.n_threads, ",").c_str());
printf(" -ngl, --n-gpu-layers <n> (default: %s)\n", join(cmd_params_defaults.n_gpu_layers, ",").c_str());
printf(" -rpc, --rpc <rpc_servers> (default: %s)\n", join(cmd_params_defaults.rpc_servers, ",").c_str());
printf(" -sm, --split-mode <none|layer|row> (default: %s)\n", join(transform_to_str(cmd_params_defaults.split_mode, split_mode_str), ",").c_str());
printf(" -mg, --main-gpu <i> (default: %s)\n", join(cmd_params_defaults.main_gpu, ",").c_str());
printf(" -nkvo, --no-kv-offload <0|1> (default: %s)\n", join(cmd_params_defaults.no_kv_offload, ",").c_str());
Expand Down Expand Up @@ -384,6 +387,12 @@ static cmd_params parse_cmd_params(int argc, char ** argv) {
}
auto p = split<int>(argv[i], split_delim);
params.n_gpu_layers.insert(params.n_gpu_layers.end(), p.begin(), p.end());
} else if (arg == "-rpc" || arg == "--rpc") {
if (++i >= argc) {
invalid_param = true;
break;
}
params.rpc_servers.push_back(argv[i]);
} else if (arg == "-sm" || arg == "--split-mode") {
if (++i >= argc) {
invalid_param = true;
Expand Down Expand Up @@ -519,6 +528,7 @@ static cmd_params parse_cmd_params(int argc, char ** argv) {
if (params.type_k.empty()) { params.type_k = cmd_params_defaults.type_k; }
if (params.type_v.empty()) { params.type_v = cmd_params_defaults.type_v; }
if (params.n_gpu_layers.empty()) { params.n_gpu_layers = cmd_params_defaults.n_gpu_layers; }
if (params.rpc_servers.empty()) { params.rpc_servers = cmd_params_defaults.rpc_servers; }
if (params.split_mode.empty()) { params.split_mode = cmd_params_defaults.split_mode; }
if (params.main_gpu.empty()) { params.main_gpu = cmd_params_defaults.main_gpu; }
if (params.no_kv_offload.empty()){ params.no_kv_offload = cmd_params_defaults.no_kv_offload; }
Expand All @@ -541,6 +551,7 @@ struct cmd_params_instance {
ggml_type type_v;
int n_threads;
int n_gpu_layers;
std::string rpc_servers;
llama_split_mode split_mode;
int main_gpu;
bool no_kv_offload;
Expand All @@ -553,6 +564,9 @@ struct cmd_params_instance {
llama_model_params mparams = llama_model_default_params();

mparams.n_gpu_layers = n_gpu_layers;
if (!rpc_servers.empty()) {
mparams.rpc_servers = rpc_servers.c_str();
}
mparams.split_mode = split_mode;
mparams.main_gpu = main_gpu;
mparams.tensor_split = tensor_split.data();
Expand All @@ -564,6 +578,7 @@ struct cmd_params_instance {
bool equal_mparams(const cmd_params_instance & other) const {
return model == other.model &&
n_gpu_layers == other.n_gpu_layers &&
rpc_servers == other.rpc_servers &&
split_mode == other.split_mode &&
main_gpu == other.main_gpu &&
use_mmap == other.use_mmap &&
Expand Down Expand Up @@ -592,6 +607,7 @@ static std::vector<cmd_params_instance> get_cmd_params_instances(const cmd_param
// this ordering minimizes the number of times that each model needs to be reloaded
for (const auto & m : params.model)
for (const auto & nl : params.n_gpu_layers)
for (const auto & rpc : params.rpc_servers)
for (const auto & sm : params.split_mode)
for (const auto & mg : params.main_gpu)
for (const auto & ts : params.tensor_split)
Expand All @@ -618,6 +634,7 @@ static std::vector<cmd_params_instance> get_cmd_params_instances(const cmd_param
/* .type_v = */ tv,
/* .n_threads = */ nt,
/* .n_gpu_layers = */ nl,
/* .rpc_servers = */ rpc,
/* .split_mode = */ sm,
/* .main_gpu = */ mg,
/* .no_kv_offload= */ nkvo,
Expand All @@ -643,6 +660,7 @@ static std::vector<cmd_params_instance> get_cmd_params_instances(const cmd_param
/* .type_v = */ tv,
/* .n_threads = */ nt,
/* .n_gpu_layers = */ nl,
/* .rpc_servers = */ rpc,
/* .split_mode = */ sm,
/* .main_gpu = */ mg,
/* .no_kv_offload= */ nkvo,
Expand All @@ -668,6 +686,7 @@ static std::vector<cmd_params_instance> get_cmd_params_instances(const cmd_param
/* .type_v = */ tv,
/* .n_threads = */ nt,
/* .n_gpu_layers = */ nl,
/* .rpc_servers = */ rpc,
/* .split_mode = */ sm,
/* .main_gpu = */ mg,
/* .no_kv_offload= */ nkvo,
Expand All @@ -692,6 +711,7 @@ struct test {
static const bool kompute;
static const bool metal;
static const bool sycl;
static const bool rpc;
static const bool gpu_blas;
static const bool blas;
static const std::string cpu_info;
Expand Down Expand Up @@ -790,6 +810,9 @@ struct test {
if (sycl) {
return GGML_SYCL_NAME;
}
if (rpc) {
return "RPC";
}
if (gpu_blas) {
return "GPU BLAS";
}
Expand All @@ -803,7 +826,7 @@ struct test {
static const std::vector<std::string> & get_fields() {
static const std::vector<std::string> fields = {
"build_commit", "build_number",
"cuda", "opencl", "vulkan", "kompute", "metal", "sycl", "gpu_blas", "blas",
"cuda", "opencl", "vulkan", "kompute", "metal", "sycl", "rpc", "gpu_blas", "blas",
"cpu_info", "gpu_info",
"model_filename", "model_type", "model_size", "model_n_params",
"n_batch", "n_ubatch",
Expand Down Expand Up @@ -859,7 +882,7 @@ struct test {
std::vector<std::string> values = {
build_commit, std::to_string(build_number),
std::to_string(cuda), std::to_string(opencl), std::to_string(vulkan), std::to_string(vulkan),
std::to_string(metal), std::to_string(sycl), std::to_string(gpu_blas), std::to_string(blas),
std::to_string(metal), std::to_string(sycl), std::to_string(rpc), std::to_string(gpu_blas), std::to_string(blas),
cpu_info, gpu_info,
model_filename, model_type, std::to_string(model_size), std::to_string(model_n_params),
std::to_string(n_batch), std::to_string(n_ubatch),
Expand Down Expand Up @@ -894,6 +917,7 @@ const bool test::metal = !!ggml_cpu_has_metal();
const bool test::gpu_blas = !!ggml_cpu_has_gpublas();
const bool test::blas = !!ggml_cpu_has_blas();
const bool test::sycl = !!ggml_cpu_has_sycl();
const bool test::rpc = !!ggml_cpu_has_rpc();
const std::string test::cpu_info = get_cpu_info();
const std::string test::gpu_info = get_gpu_info();

Expand Down
8 changes: 8 additions & 0 deletions ggml.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -22872,6 +22872,14 @@ int ggml_cpu_has_sycl(void) {
#endif
}

int ggml_cpu_has_rpc(void) {
#if defined(GGML_USE_RPC)
return 1;
#else
return 0;
#endif
}

int ggml_cpu_has_gpublas(void) {
return ggml_cpu_has_cuda() || ggml_cpu_has_clblast() || ggml_cpu_has_vulkan() || ggml_cpu_has_kompute() ||
ggml_cpu_has_sycl();
Expand Down
1 change: 1 addition & 0 deletions ggml.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2428,6 +2428,7 @@ extern "C" {
GGML_API int ggml_cpu_has_sse3 (void);
GGML_API int ggml_cpu_has_ssse3 (void);
GGML_API int ggml_cpu_has_sycl (void);
GGML_API int ggml_cpu_has_rpc (void);
GGML_API int ggml_cpu_has_vsx (void);
GGML_API int ggml_cpu_has_matmul_int8(void);

Expand Down

0 comments on commit 210d991

Please sign in to comment.