tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-29 05:33:37 -04:00
Code Issues Packages Projects Releases Wiki Activity

kv-cache : refactor the update/defrag mechanism (#13988)

Browse Source 
* kv-cache : refactor update mechanism

ggml-ci

* memory : improve status handling

* defrag : reset head + add comments

ggml-ci

* cont : minor fixes

ggml-ci
This commit is contained in:
Georgi Gerganov
2025-06-04 18:58:20 +03:00
committed by GitHub
parent 2589ad3704
commit 3e63a58ef7
11 changed files with 340 additions and 191 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
src/llama-kv-cache-unified-iswa.h
View File

@@ -54,9 +54,7 @@ public:
llama_memory_state_ptr init_full() override;
bool update(llama_context & lctx) override;
void defrag_sched(float thold) override;
llama_memory_state_ptr init_update(llama_context * lctx, bool optimize) override;
bool get_can_shift() const override;
@@ -86,12 +84,16 @@ public:
// used to create a full-cache state
llama_kv_cache_unified_iswa_state(
llama_memory_status status,
llama_kv_cache_unified_iswa * kv);
// used to create an update state
llama_kv_cache_unified_iswa_state(
llama_kv_cache_unified_iswa * kv,
llama_context * lctx,
bool optimize);
// used to create a state from a batch
llama_kv_cache_unified_iswa_state(
llama_memory_status status,
llama_kv_cache_unified_iswa * kv,
llama_sbatch sbatch,
std::vector<uint32_t> heads_base,
@@ -120,7 +122,7 @@ public:
const llama_kv_cache_unified_state * get_swa() const;
private:
const llama_memory_status status;
llama_memory_status status;
//llama_kv_cache_unified_iswa * kv;
@@ -131,6 +133,6 @@ private:
std::vector<llama_ubatch> ubatches;
std::unique_ptr<llama_kv_cache_unified_state> state_base;
std::unique_ptr<llama_kv_cache_unified_state> state_swa;
llama_memory_state_ptr state_base;
llama_memory_state_ptr state_swa;
};