Add qnn_quantized_test

test/CMakeLists.txt

@@ -53,7 +53,7 @@ endif()
 
 if(edgerunner_ENABLE_NPU)
     list(APPEND TEST_SOURCES source/qnn_shared_library_npu_test.cpp
-         source/qnn_context_binary_npu_test.cpp
+         source/qnn_context_binary_npu_test.cpp source/qnn_quantized_test.cpp
     )
 endif()
 

test/source/qnn_quantized_test.cpp

@@ -0,0 +1,66 @@
+#include <cstddef>
+#include <cstdint>
+#include <string>
+#include <vector>
+
+#include <catch2/catch_test_macros.hpp>
+
+#include "edgerunner/edgerunner.hpp"
+#include "edgerunner/model.hpp"
+#include "edgerunner/tensor.hpp"
+#include "utils.hpp"
+
+TEST_CASE("QNN runtime quantized (NPU)", "[qnn][npu][quantized]") {
+    const std::string modelPath = "models/qnn/mobilenet_v3_large_quantized.so";
+
+    auto model = edge::createModel(modelPath);
+    REQUIRE(model != nullptr);
+    REQUIRE(std::string {"mobilenet_v3_large_quantized"} == model->name());
+
+    model->applyDelegate(edge::DELEGATE::CPU);
+    REQUIRE(model->getDelegate() == edge::DELEGATE::CPU);
+
+    REQUIRE(model->getPrecision() == edge::TensorType::UINT8);
+
+    const auto inputs = model->getInputs();
+    const auto numInputs = model->getNumInputs();
+    REQUIRE(numInputs == 1);
+    REQUIRE(numInputs == inputs.size());
+
+    const auto outputs = model->getOutputs();
+    const auto numOutputs = model->getNumOutputs();
+    REQUIRE(numOutputs == 1);
+    REQUIRE(numOutputs == outputs.size());
+
+    auto input = model->getInput(0);
+    REQUIRE(input->getName() == "image_tensor");
+    REQUIRE(input->getDimensions() == std::vector<size_t> {1, 224, 224, 3});
+    REQUIRE(input->getType() == edge::TensorType::UINT8);
+    REQUIRE(input.get() == inputs[0].get());
+
+    auto inputData = input->getTensorAs<uint8_t>();
+    REQUIRE(inputData.size() == input->getSize());
+
+    auto badInput = model->getInput(1);
+    REQUIRE(badInput == nullptr);
+
+    auto output = model->getOutput(0);
+    REQUIRE(output->getName() == "_668");
+    REQUIRE(output->getDimensions() == std::vector<size_t> {1, 1000});
+    REQUIRE(output->getType() == edge::TensorType::UINT8);
+    REQUIRE(output.get() == outputs[0].get());
+
+    auto outputBuffer = output->getTensorAs<uint8_t>();
+    REQUIRE(outputBuffer.size() == output->getSize());
+
+    auto badOutput = model->getOutput(1);
+    REQUIRE(badOutput == nullptr);
+
+    const auto executionStatus = model->execute();
+    REQUIRE(executionStatus == edge::STATUS::SUCCESS);
+
+    /* verify output buffer is persistent across execution */
+    const auto newOutputBuffer = model->getOutput(0)->getTensorAs<uint8_t>();
+    REQUIRE(outputBuffer.data() == newOutputBuffer.data());
+    REQUIRE(outputBuffer.size() == newOutputBuffer.size());
+}