performance test: improved macros for test setup, fixes for gcc

algorithms_cpp/perf/for_each_perf-inl.h

@@ -41,7 +41,7 @@ namespace perf {
 template<typename T>
 template<typename T>
 SerialForEach<T>::SerialForEach(const embb::base::perf::CallArgs & args)
 SerialForEach<T>::SerialForEach(const embb::base::perf::CallArgs & args)
 : cargs(args), op(args), vector_size(args.VectorSize()) {
 : cargs(args), op(args), vector_size(args.VectorSize()) {
-  if (cargs.StressMode() == CallArgs::RAM_STRESS) {
+  if (cargs.StressMode() == embb::base::perf::CallArgs::RAM_STRESS) {
     v = static_cast<T *>(embb::base::Allocation::AllocateCacheAligned(
     v = static_cast<T *>(embb::base::Allocation::AllocateCacheAligned(
       vector_size * sizeof(T)));
       vector_size * sizeof(T)));
     for (size_t i = 0; i < vector_size; i++) {
     for (size_t i = 0; i < vector_size; i++) {
@@ -54,19 +54,19 @@ SerialForEach<T>::SerialForEach(const embb::base::perf::CallArgs & args)
 
 
 template<typename T>
 template<typename T>
 SerialForEach<T>::~SerialForEach() {
 SerialForEach<T>::~SerialForEach() {
-  if (cargs.StressMode() == CallArgs::RAM_STRESS) {
+  if (cargs.StressMode() == embb::base::perf::CallArgs::RAM_STRESS) {
     embb::base::Allocation::FreeAligned(v);
     embb::base::Allocation::FreeAligned(v);
   }
   }
 }
 }
 
 
 template<typename T>
 template<typename T>
 void SerialForEach<T>::Run() {
 void SerialForEach<T>::Run() {
-  if (cargs.StressMode() == CallArgs::CPU_STRESS) {
+  if (cargs.StressMode() == embb::base::perf::CallArgs::CPU_STRESS) {
     for (size_t i = 0; i < vector_size; i++) {
     for (size_t i = 0; i < vector_size; i++) {
       T v = static_cast<T>(i);
       T v = static_cast<T>(i);
       op(v);
       op(v);
     }
     }
-  } else if (cargs.StressMode() == CallArgs::RAM_STRESS) {
+  } else if (cargs.StressMode() == embb::base::perf::CallArgs::RAM_STRESS) {
     for (size_t i = 0; i < vector_size; i++) {
     for (size_t i = 0; i < vector_size; i++) {
       op(v[i]);
       op(v[i]);
     }
     }
@@ -76,7 +76,7 @@ void SerialForEach<T>::Run() {
 template<typename T>
 template<typename T>
 ParallelForEach<T>::ParallelForEach(const embb::base::perf::CallArgs & args)
 ParallelForEach<T>::ParallelForEach(const embb::base::perf::CallArgs & args)
 : cargs(args), vector_size(args.VectorSize()) {
 : cargs(args), vector_size(args.VectorSize()) {
-  if (cargs.StressMode() == CallArgs::RAM_STRESS) {
+  if (cargs.StressMode() == embb::base::perf::CallArgs::RAM_STRESS) {
     v = static_cast<T *>(embb::base::Allocation::AllocateCacheAligned(
     v = static_cast<T *>(embb::base::Allocation::AllocateCacheAligned(
       vector_size * sizeof(T)));
       vector_size * sizeof(T)));
   } else {
   } else {
@@ -93,7 +93,7 @@ ParallelForEach<T>::~ParallelForEach() {
 
 
 template<typename T>
 template<typename T>
 void ParallelForEach<T>::Pre() {
 void ParallelForEach<T>::Pre() {
-  if (cargs.StressMode() == CallArgs::RAM_STRESS) {
+  if (cargs.StressMode() == embb::base::perf::CallArgs::RAM_STRESS) {
     // Initialize input vector with incrementing values:
     // Initialize input vector with incrementing values:
     for (size_t i = 0; i < vector_size; i++) {
     for (size_t i = 0; i < vector_size; i++) {
       v[i] = static_cast<T>(i);
       v[i] = static_cast<T>(i);
@@ -103,7 +103,7 @@ void ParallelForEach<T>::Pre() {
 
 
 template<typename T>
 template<typename T>
 void ParallelForEach<T>::Run(unsigned int numThreads) {
 void ParallelForEach<T>::Run(unsigned int numThreads) {
-  if (cargs.StressMode() == CallArgs::CPU_STRESS) {
+  if (cargs.StressMode() == embb::base::perf::CallArgs::CPU_STRESS) {
     // Computing input values, no memory access
     // Computing input values, no memory access
     ForEachFunctor<T> op(cargs);
     ForEachFunctor<T> op(cargs);
     embb::algorithms::ForEach(
     embb::algorithms::ForEach(
@@ -114,7 +114,7 @@ void ParallelForEach<T>::Run(unsigned int numThreads) {
       op,
       op,
       embb::tasks::ExecutionPolicy(),
       embb::tasks::ExecutionPolicy(),
       vector_size / numThreads);
       vector_size / numThreads);
-  } else if (cargs.StressMode() == CallArgs::RAM_STRESS) {
+  } else if (cargs.StressMode() == embb::base::perf::CallArgs::RAM_STRESS) {
     // Reading input values from memory
     // Reading input values from memory
     ForEachFunctor<T> op(cargs);
     ForEachFunctor<T> op(cargs);
     embb::algorithms::ForEach(
     embb::algorithms::ForEach(

base_cpp/perf/include/embb/base/perf/perf.h

@@ -34,11 +34,6 @@
 #include <iomanip>
 #include <iomanip>
 
 
 #define PT_PERF_MAIN(component) \
 #define PT_PERF_MAIN(component) \
-template <class Test> \
-void PartestRunPerformanceTest(Test & test) { \
-  test.Run(); \
-  test.PrintReport(std::cout); \
-} \
 void PartestRunPerformanceTests( \
 void PartestRunPerformanceTests( \
   embb::base::perf::CallArgs & perf_test_params); \
   embb::base::perf::CallArgs & perf_test_params); \
 int main(int argc, char** argv) { \
 int main(int argc, char** argv) { \
@@ -59,7 +54,11 @@ void PartestRunPerformanceTests( \
 #define PT_PERF_RUN(PT_PERF_TEST) \
 #define PT_PERF_RUN(PT_PERF_TEST) \
 ( \
 ( \
   (std::cout << "Running " << #PT_PERF_TEST << " ..." << std::endl), \
   (std::cout << "Running " << #PT_PERF_TEST << " ..." << std::endl), \
-  PartestRunPerformanceTest<PT_PERF_TEST>(PT_PERF_TEST(perf_test_params)), \
+  ({ \
+    PT_PERF_TEST perf_test(perf_test_params); \
+    perf_test.Run(); \
+    perf_test.PrintReport(std::cout); \
+  }), \
   (std::cout << "Running " << #PT_PERF_TEST << " ..." << " done" << std::endl) \
   (std::cout << "Running " << #PT_PERF_TEST << " ..." << " done" << std::endl) \
 )
 )