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

llama_model_loader: only map tensors included in the context

Browse source
This commit is contained in:
Pierrick HYMBERT 2024-03-21 21:42:30 +01:00
parent 078a1aca06
commit 69bdee939a
1 changed files with 0 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

1
llama.cpp
View file

@ -5171,7 +5171,6 @@ static bool llm_load_tensors(
size_t first, last; size_t first, last;
ml.get_mapping_range(&first, &last, &addr, file_no, ctx); ml.get_mapping_range(&first, &last, &addr, file_no, ctx);
if (first >= last) { if (first >= last) {
bufs.push_back(nullptr); // add a dummy buffer to keep the indices in sync
continue; continue;
} }
ggml_backend_buffer_t buf = ggml_backend_metal_buffer_from_ptr((char *) addr + first, last - first, max_size); ggml_backend_buffer_t buf = ggml_backend_metal_buffer_from_ptr((char *) addr + first, last - first, max_size);