ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

memory : rename interface to llama_memory_context_i (#14296)

Browse source 
* memory : rename interface to llama_memory_context_i

ggml-ci

* cont : fix comments

* cont : use "mctx" for referencing a memory context

ggml-ci
This commit is contained in:
Georgi Gerganov 2025-06-21 08:03:46 +03:00 • committed by GitHub
parent b23fa0b3f4
commit 692e3cdd0a
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
14 changed files with 339 additions and 341 deletions
Download patch file Download diff file Expand all files Collapse all files

32
src/llama-kv-cache-unified.h
View file

@ -56,14 +56,14 @@ public:
// llama_memory_i
//
llama_memory_state_ptr init_batch(
llama_memory_context_ptr init_batch(
llama_batch_allocr & balloc,
uint32_t n_ubatch,
bool embd_all) override;
llama_memory_state_ptr init_full() override;
llama_memory_context_ptr init_full() override;
llama_memory_state_ptr init_update(llama_context * lctx, bool optimize) override;
llama_memory_context_ptr init_update(llama_context * lctx, bool optimize) override;
bool get_can_shift() const override;
@ -208,36 +208,36 @@ private:
bool state_read_data(llama_io_read_i & io, uint32_t cell_count);
};
class llama_kv_cache_unified_state : public llama_memory_state_i {
class llama_kv_cache_unified_context : public llama_memory_context_i {
public:
// some shorthands
using ubatch_heads = llama_kv_cache_unified::ubatch_heads;
using defrag_info = llama_kv_cache_unified::defrag_info;
// used for errors
llama_kv_cache_unified_state(llama_memory_status status);
llama_kv_cache_unified_context(llama_memory_status status);
// used to create a full-cache state
llama_kv_cache_unified_state(
// used to create a full-cache context
llama_kv_cache_unified_context(
llama_kv_cache_unified * kv);
// used to create an update state
llama_kv_cache_unified_state(
// used to create an update context
llama_kv_cache_unified_context(
llama_kv_cache_unified * kv,
llama_context * lctx,
bool do_shift,
defrag_info dinfo);
// used to create a decode state from a batch
llama_kv_cache_unified_state(
// used to create a batch procesing context from a batch
llama_kv_cache_unified_context(
llama_kv_cache_unified * kv,
ubatch_heads heads,
std::vector<llama_ubatch> ubatches);
virtual ~llama_kv_cache_unified_state();
virtual ~llama_kv_cache_unified_context();
//
// llama_memory_state_i
// llama_memory_context_i
//
bool next() override;
@ -247,7 +247,7 @@ public:
const llama_ubatch & get_ubatch() const override;
//
// llama_kv_cache_unified_state specific API
// llama_kv_cache_unified_context specific API
//
uint32_t get_n_kv() const;
@ -272,7 +272,7 @@ private:
llama_context * lctx;
//
// update state
// update context
//
bool do_shift = false;
@ -280,7 +280,7 @@ private:
defrag_info dinfo;
//
// batch processing state
// batch processing context
//
// the index of the next ubatch to process