vrr/prima.cpp
2
0
Fork 0
mirror of https://github.com/Lizonghang/prima.cpp.git synced 2025-09-09 00:29:07 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

use user input setup in standalone mode

Browse source
This commit is contained in:
Lizonghang 2025-01-21 21:06:09 +04:00
parent 0b748060ad
commit 871a27f66a
1 changed files with 8 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
common/common.cpp
View file

@ -1045,6 +1045,7 @@ static void assign_device(
printf("\n"); printf("\n");
} }
}; };
(void)print_matrix;
double final_objective = 1.0e30; double final_objective = 1.0e30;
std::vector<double> final_solution; std::vector<double> final_solution;
@ -1412,7 +1413,7 @@ struct llama_init_result llama_init_from_gpt_params(gpt_params & params) {
device_info dev_info; device_info dev_info;
uint32_t n_world = params.n_world; uint32_t n_world = params.n_world;
uint32_t my_rank = params.rank; uint32_t my_rank = params.rank;
bool auto_schedule = n_world == 1 || params.n_layer_window[0] == 0; bool auto_schedule = n_world > 1 && params.n_layer_window[0] == 0;
if (auto_schedule) { if (auto_schedule) {
// get device profile // get device profile
@ -1460,6 +1461,12 @@ struct llama_init_result llama_init_from_gpt_params(gpt_params & params) {
cparams.n_gpu_layers = n_gpu_layers[my_rank]; cparams.n_gpu_layers = n_gpu_layers[my_rank];
mparams.n_gpu_layers = n_gpu_layers[my_rank]; mparams.n_gpu_layers = n_gpu_layers[my_rank];
llama_model_set_n_gpu_layers(model, n_gpu_layers[my_rank]); llama_model_set_n_gpu_layers(model, n_gpu_layers[my_rank]);
} else if (n_world == 1) {
uint32_t n_layers = llama_model_n_layers(model);
params.n_layer_window[0] = n_layers;
cparams.n_layer_window[0] = n_layers;
mparams.n_layer_window[0] = n_layers;
llama_context_n_layer_window(lctx)[0] = n_layers;
} }
LOG_INF("\nUsing window size: %d, GPU layers: %d\n\n", cparams.n_layer_window[my_rank], cparams.n_gpu_layers); LOG_INF("\nUsing window size: %d, GPU layers: %d\n\n", cparams.n_layer_window[my_rank], cparams.n_gpu_layers);