EngineX-Ascend/enginex-ascend-910-llama.cpp
10
0
Fork 0
Code Issues Pull Requests Actions 4 Projects Releases Wiki Activity

vulkan: sort graph to allow more parallel execution (#15850)

Browse Source 
* vulkan: sort graph to allow more parallel execution

Add a backend proc to allow the backend to modify the graph. The
vulkan implementation looks at which nodes depend on each other
and greedily reorders them to group together nodes that don't
depend on each other. It only reorders the nodes, doesn't change
the contents of any of them.

With #15489, this reduces the number of synchronizations needed.

* call optimize_graph per-split
This commit is contained in:
Jeff Bolz
2025-09-08 13:10:07 -05:00
committed by GitHub
parent 0a16bf52e6
commit e68aa10d8f
13 changed files with 154 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
ggml/src/ggml-backend-impl.h
View File

@@ -114,6 +114,9 @@ extern "C" {
void (*event_record)(ggml_backend_t backend, ggml_backend_event_t event);
// wait for an event on on a different stream
void (*event_wait) (ggml_backend_t backend, ggml_backend_event_t event);
// (optional) sort/optimize the nodes in the graph
void (*optimize_graph) (ggml_backend_t backend, struct ggml_cgraph * cgraph);
};
struct ggml_backend {