perf: reduce CPU overhead in graph execution and sampling (#1997)

This commit is contained in:
leejet authored and GitHub committed 2026-09-19 18:32:58 +08:00
1 parent d32b4e893b
commit 275ab58e01
8 files changed
+195 -123

No files matched your search

+2 -4
View File
@@ -265,11 +265,9 @@ protected:
void copy_data_to_backend_tensor(ggml_cgraph* gf, bool clear_after_copy = true);
bool resolve_graph_cut_plan(ggml_cgraph* gf,
GraphCutPlan* plan_out);
const GraphCutPlan& resolve_graph_cut_plan(ggml_cgraph* gf);
bool resolve_graph_cut_layer_split_plan(ggml_cgraph* gf,
GraphCutPlan* plan_out);
const GraphCutPlan& resolve_graph_cut_layer_split_plan(ggml_cgraph* gf);
bool assign_graph_cut_layer_split_backends(ggml_cgraph* gf);