Remove map/unmap overhead for input/output accessor when using DummyAccessor

Don't map/unmap when the tensor data is not accessed in the input or output accessor.
This is so to avoid measuring CPU overhead when benchmarking on the GPU backend.

Resolve COMPMID-4712

Change-Id: I5baba1b93e7a51fe13525bcce6c0cfdecb14493e
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6140
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

1 files changed, 9 insertions, 1 deletions

diff --git a/arm_compute/graph/ITensorAccessor.h b/arm_compute/graph/ITensorAccessor.h
index 96bd499573..a8818be2e5 100644
--- a/arm_compute/graph/ITensorAccessor.h
+++ b/arm_compute/graph/ITensorAccessor.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018-2019 Arm Limited.
+ * Copyright (c) 2018-2019,2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -45,6 +45,14 @@ public:
      * @return True if access is successful else false
      */
     virtual bool access_tensor(ITensor &tensor) = 0;
+    /** Returns true if the tensor data is being accessed
+     *
+     * @return True if the tensor data is being accessed by the accessor. False otherwise
+     */
+    virtual bool access_tensor_data()
+    {
+        return true;
+    }
 };
 
 using ITensorAccessorUPtr = std::unique_ptr<ITensorAccessor>;