CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-02 09:12:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

allocators wip

Browse Source 
renamed ggml_backend functions
changed ggml_buffer and ggml_backend to always be used as pointers
rename ggml_tensor::params -> op_params
This commit is contained in:
slaren
2023-07-17 19:03:51 +02:00
parent 1102ff56db
commit 295f85654a
8 changed files with 640 additions and 365 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
ggml.h
View File

@@ -414,19 +414,21 @@ extern "C" {
// compute data
enum ggml_op op;
// op params - allocated as int32_t for alignment
int32_t op_params[GGML_MAX_OP_PARAMS / sizeof(uint32_t)];
bool is_param;
struct ggml_tensor * grad;
struct ggml_tensor * src[GGML_MAX_SRC];
int node_id; // used to build graphs
// performance
int perf_runs;
int64_t perf_cycles;
int64_t perf_time_us;
// op params
// allocated as int32_t to avoid alignment issues
int32_t params[GGML_MAX_OP_PARAMS / sizeof(uint32_t)];
void * data;
@@ -434,7 +436,6 @@ extern "C" {
void * extra; // extra things e.g. for ggml-cuda.cu
bool visited; // used to build graphs
char padding[4];
};