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

add tool to allow plotting tensor allocation maps within buffers

Browse source
This commit is contained in:
slaren 2024-08-06 22:09:51 +02:00
parent 5f4dcb1e60
commit cad8abb49b
6 changed files with 336 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

3
ggml/include/ggml-alloc.h
View file

@ -71,6 +71,9 @@ GGML_API size_t ggml_gallocr_get_buffer_size(ggml_gallocr_t galloc, int buffer_i
GGML_API struct ggml_backend_buffer * ggml_backend_alloc_ctx_tensors_from_buft(struct ggml_context * ctx, ggml_backend_buffer_type_t buft);
GGML_API struct ggml_backend_buffer * ggml_backend_alloc_ctx_tensors(struct ggml_context * ctx, ggml_backend_t backend);
// Export tensor allocations in a graph to a file that can be plotted
GGML_API void ggml_gallocr_export_allocs(const char * filename, struct ggml_cgraph * graph);
#ifdef __cplusplus
}
#endif

3
ggml/include/ggml-backend.h
View file

@ -208,6 +208,9 @@ extern "C" {
// Set a callback to be called for each resulting node during graph compute
GGML_API void ggml_backend_sched_set_eval_callback(ggml_backend_sched_t sched, ggml_backend_sched_eval_callback callback, void * user_data);
// internal
GGML_API struct ggml_cgraph * ggml_backend_sched_get_graph_copy(ggml_backend_sched_t sched);
//
// Utils
//