Qualcomm AI Engine Direct - Optimize memory usage at runtime (#7003)

shewu-quic · web-flow · commit 96a9d35f7c5b · 2024-11-21T14:23:43.000-08:00
Qualcomm AI Engine Direct - Optimize memory at runtime
diff --git a/backends/qualcomm/runtime/QnnExecuTorchBackend.cpp b/backends/qualcomm/runtime/QnnExecuTorchBackend.cpp
@@ -90,6 +90,8 @@ Result<DelegateHandle*> QnnExecuTorchBackend::init(
     }
   }
   add_cached_delegate(signature, qnn_manager);
+  // This backend does not need its processed data after Init.
+  processed->Free();
   return qnn_manager;
 }
 

Original file line number	Diff line number	Diff line change
`@@ -90,6 +90,8 @@ Result<DelegateHandle*> QnnExecuTorchBackend::init(`
`90`	`90`	`}`
`91`	`91`	`}`
`92`	`92`	`add_cached_delegate(signature, qnn_manager);`
	`93`	`+ // This backend does not need its processed data after Init.`
	`94`	`+ processed->Free();`
`93`	`95`	`return qnn_manager;`
`94`	`96`	`}`
`95`	`97`