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

ggml : better PERF prints + support "LLAMA_PERF=1 make"

Browse source
This commit is contained in:
Georgi Gerganov 2023-04-23 18:15:39 +03:00
parent 53c8434398
commit e4422e299c
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
3 changed files with 9 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

4
llama.cpp
View file

@ -1250,9 +1250,11 @@ static bool llama_eval_internal(
ggml_build_forward_expand(&gf, inpL);
ggml_graph_compute (ctx0, &gf);
#ifdef GGML_PERF
// print timing information per ggml operation (for debugging purposes)
// requires GGML_PERF to be defined
//ggml_graph_print(&gf);
ggml_graph_print(&gf);
#endif
// plot the computation graph in dot format (for debugging purposes)
//if (n_past%100 == 0) {