CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-01 09:01:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama: print memory breakdown on exit (#15860)

Browse Source 
* llama: print memory breakdown on exit
This commit is contained in:
Johannes Gäßler
2025-09-24 16:53:48 +02:00
committed by GitHub
parent f2a789e334
commit e789095502
18 changed files with 243 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/llama-kv-cache-iswa.cpp
View File

@@ -113,6 +113,14 @@ llama_pos llama_kv_cache_iswa::seq_pos_max(llama_seq_id seq_id) const {
return kv_swa->seq_pos_max(seq_id);
}
std::map<ggml_backend_buffer_type_t, size_t> llama_kv_cache_iswa::memory_breakdown() const {
std::map<ggml_backend_buffer_type_t, size_t> mb = kv_base->memory_breakdown();
for (const auto & buft_size : kv_swa->memory_breakdown()) {
mb[buft_size.first] += buft_size.second;
}
return mb;
}
llama_memory_context_ptr llama_kv_cache_iswa::init_batch(llama_batch_allocr & balloc, uint32_t n_ubatch, bool embd_all) {
GGML_UNUSED(embd_all);