commit 9b8e4c60117dcea604d35444c99c525cafca43d3
author Keith Mok <ek9852@gmail.com> Mon Mar 01 11:40:18 2021 -0800
committer mike.kelly <mike.kelly@arm.com> Thu Mar 04 16:30:37 2021 +0000
tree 79cc698c205cb4781222bdeae22467fa6fbd5376
parent f806c4d075814a9dc9d206a4db123d3060ad7ebd

armnn_tfl_benchmark: allow string of backend to pass directly

Instead of using the predefined backend ids,
we should just pass the backend string directly to the optimator,
otherwise we cannot use vendor specific backends or GpuAcc

Signed-off-by: Keith Mok <ek9852@gmail.com>
Change-Id: Ic52c81d48364a19f0fcc59bc3c975cf3bd740e21

tests/TfLiteBenchmark-Armnn/TfLiteBenchmark-Armnn.cpp

diff --git a/tests/TfLiteBenchmark-Armnn/TfLiteBenchmark-Armnn.cpp b/tests/TfLiteBenchmark-Armnn/TfLiteBenchmark-Armnn.cpp
index 4e092cf..3046e7e 100644
--- a/tests/TfLiteBenchmark-Armnn/TfLiteBenchmark-Armnn.cpp
+++ b/tests/TfLiteBenchmark-Armnn/TfLiteBenchmark-Armnn.cpp
@@ -18,7 +18,8 @@
 #include <armnnTfLiteParser/ITfLiteParser.hpp>
 
 // Application parameters
-std::vector<armnn::BackendId> preferred_backends_order = {armnn::Compute::CpuAcc, armnn::Compute::CpuRef};
+std::vector<armnn::BackendId> default_preferred_backends_order = {armnn::Compute::CpuAcc, armnn::Compute::CpuRef};
+std::vector<armnn::BackendId> preferred_backends_order;
 std::string model_file_str;
 std::string preferred_backend_str;
 int nb_loops = 1;
@@ -45,7 +46,8 @@
         "-m --model_file <.tflite file path>:  .tflite model to be executed\n"
         "-b --backend <device>:                preferred backend device to run layers on by default. Possible choices: "
                                                << armnn::BackendRegistryInstance().GetBackendIdsAsString() << "\n"
-        "-l --loops <int>:                     provide the number of time the inference will be executed\n"
+        "                                      (by default CpuAcc, CpuRef)\n"
+        "-l --loops <int>:                     provide the number of times the inference will be executed\n"
         "                                      (by default nb_loops=1)\n"
         "--help:                               show this help\n";
     exit(1);
@@ -79,22 +81,10 @@
             break;
         case 'b':
             preferred_backend_str = std::string(optarg);
-            // Overwrite the prefered backend order
-            if (preferred_backend_str == "CpuAcc")
-            {
-                preferred_backends_order = {armnn::Compute::CpuAcc, armnn::Compute::CpuRef};
-            }
-            else if (preferred_backend_str == "CpuRef")
-            {
-                preferred_backends_order = {armnn::Compute::CpuRef, armnn::Compute::CpuAcc};
-            }
+            // Overwrite the backend
+            preferred_backends_order.push_back(preferred_backend_str);
 
-            std::cout << "preferred backend device set to:";
-            for (unsigned int i = 0; i < preferred_backends_order.size(); i++)
-            {
-                std::cout << " " << preferred_backends_order.at(i);
-            }
-            std::cout << std::endl;
+            std::cout << "backend device set to:" << preferred_backend_str << std::endl;;
             break;
         case 'l':
             nb_loops = std::stoi(optarg);
@@ -136,6 +126,10 @@
     }
 
     // Optimize the network
+    if (preferred_backends_order.size() == 0)
+    {
+        preferred_backends_order = default_preferred_backends_order;
+    }
     armnn::IOptimizedNetworkPtr optimizedNet = armnn::Optimize(*network,
                                                                preferred_backends_order,
                                                                runtime->GetDeviceSpec());