happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Fix: For pipeline parallelism, adds missing synchronization between input copying and graph computation.

This commit is contained in:
aendk 2026-03-19 14:54:34 +01:00
parent 5e5385d0e4
commit a48fd3bedc
1 changed files with 1 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ggml/src/ggml-backend.cpp
View File

@ -1578,9 +1578,7 @@ static enum ggml_status ggml_backend_sched_compute_splits(ggml_backend_sched_t s
}
}
if (sched->events[split_backend_id][sched->cur_copy] == NULL) {
ggml_backend_synchronize(split_backend);
}
ggml_backend_synchronize(split_backend);
if (!sched->callback_eval) {
enum ggml_status ec = ggml_backend_graph_compute_async(split_backend, &split->graph);