modify default thread num

mindspore/lite/java/src/test/java/com/mindspore/ModelTest.java

@@ -61,6 +61,7 @@ public class ModelTest {
         Graph g = new Graph();
         assertTrue(g.load("../test/ut/src/runtime/kernel/arm/test_data/nets/lenet_train.ms"));
         MSContext context = new MSContext();
+        context.init(1, 0);
         TrainCfg cfg = new TrainCfg();
         Model liteModel = new Model();
         boolean isSuccess = liteModel.build(g, context, cfg);
@@ -74,7 +75,7 @@ public class ModelTest {
         Graph g = new Graph();
         assertTrue(g.load(modelFile));
         MSContext context = new MSContext();
-        context.init();
+        context.init(1,0);
         context.addDeviceInfo(DeviceType.DT_CPU, false, 0);
         Model liteModel = new Model();
         boolean isSuccess = liteModel.build(g, context, null);
@@ -119,6 +120,7 @@ public class ModelTest {
     public void testBuildByFileFailed() {
         String modelFile = "../test/ut/src/runtime/kernel/arm/test_data/nets/lenet_tod_infer.ms";
         MSContext context = new MSContext();
+        context.init(1, 0);
         Model liteModel = new Model();
         boolean isSuccess = liteModel.build(modelFile, 0, context);
         assertFalse(isSuccess);

mindspore/lite/src/runtime/cxx_api/context.h

@@ -33,7 +33,7 @@ struct Context::Data {
   std::vector<std::shared_ptr<DeviceInfoContext>> device_info_list;
 
 #ifdef PARALLEL_INFERENCE
-  int32_t thread_num = 8;
+  int32_t thread_num = 0;  // defaults are automatically adjusted based on computer performance
   int affinity_mode_ = 1;
   int32_t inter_op_parallel_num_ = 4;
 #else

mindspore/lite/src/runtime/cxx_api/model_pool/model_pool.cc

@@ -296,10 +296,19 @@ std::shared_ptr<Context> ModelPool::GetUserDefineContext(const std::shared_ptr<R
     MS_LOG(ERROR) << "user set config context nullptr.";
     return nullptr;
   }
-  if (context->GetThreadNum() <= 0) {
+  if (context->GetThreadNum() < 0) {
     MS_LOG(ERROR) << "Invalid thread num " << context->GetThreadNum();
     return nullptr;
   }
+  if (context->GetThreadNum() == 0) {
+    // Defaults are automatically adjusted based on computer performance
+    auto thread_num = GetDefaultThreadNum();
+    if (thread_num == 0) {
+      MS_LOG(ERROR) << "computer thread num failed.";
+      return nullptr;
+    }
+    context->SetThreadNum(thread_num);
+  }
   if (!context->GetThreadAffinityCoreList().empty()) {
     MS_LOG(ERROR) << "parallel predict not support user set core list.";
     return nullptr;

mindspore/lite/test/ut/python/test_inference_api.py

@@ -116,51 +116,51 @@ def test_context_04():
 
 def test_context_05():
     with pytest.raises(TypeError) as raise_info:
-        context = mslite.Context(thread_affinity_mode="1")
+        context = mslite.Context(thread_num=2, thread_affinity_mode="1")
     assert "thread_affinity_mode must be int" in str(raise_info.value)
 
 
 def test_context_06():
-    context = mslite.Context(thread_affinity_mode=2)
+    context = mslite.Context(thread_num=2, thread_affinity_mode=2)
     assert "thread_affinity_mode: 2" in str(context)
 
 
 def test_context_07():
     with pytest.raises(TypeError) as raise_info:
-        context = mslite.Context(thread_affinity_core_list=2)
+        context = mslite.Context(thread_num=2, thread_affinity_core_list=2)
     assert "thread_affinity_core_list must be list" in str(raise_info.value)
 
 
 def test_context_08():
-    context = mslite.Context(thread_affinity_core_list=[2])
+    context = mslite.Context(thread_num=2, thread_affinity_core_list=[2])
     assert "thread_affinity_core_list: [2]" in str(context)
 
 
 def test_context_09():
     with pytest.raises(TypeError) as raise_info:
-        context = mslite.Context(thread_affinity_core_list=["1", "0"])
+        context = mslite.Context(thread_num=2, thread_affinity_core_list=["1", "0"])
     assert "thread_affinity_core_list element must be int" in str(raise_info.value)
 
 
 def test_context_10():
-    context = mslite.Context(thread_affinity_core_list=[1, 0])
+    context = mslite.Context(thread_num=2, thread_affinity_core_list=[1, 0])
     assert "thread_affinity_core_list: [1, 0]" in str(context)
 
 
 def test_context_11():
     with pytest.raises(TypeError) as raise_info:
-        context = mslite.Context(enable_parallel=1)
+        context = mslite.Context(thread_num=2, enable_parallel=1)
     assert "enable_parallel must be bool" in str(raise_info.value)
 
 
 def test_context_12():
-    context = mslite.Context(enable_parallel=True)
+    context = mslite.Context(thread_num=2, enable_parallel=True)
     assert "enable_parallel: True" in str(context)
 
 
 def test_context_13():
     with pytest.raises(TypeError) as raise_info:
-        context = mslite.Context()
+        context = mslite.Context(thread_num=2)
         context.append_device_info("CPUDeviceInfo")
     assert "device_info must be DeviceInfo" in str(raise_info.value)
 
@@ -168,7 +168,7 @@ def test_context_13():
 def test_context_14():
     gpu_device_info = mslite.GPUDeviceInfo()
     cpu_device_info = mslite.CPUDeviceInfo()
-    context = mslite.Context()
+    context = mslite.Context(thread_num=2)
     context.append_device_info(gpu_device_info)
     context.append_device_info(cpu_device_info)
     assert "device_list: 1, 0" in str(context)
@@ -293,7 +293,7 @@ def test_model_01():
 
 def test_model_build_01():
     with pytest.raises(TypeError) as raise_info:
-        context = mslite.Context()
+        context = mslite.Context(thread_num=2)
         model = mslite.Model()
         model.build_from_file(model_path=1, model_type=mslite.ModelType.MINDIR_LITE, context=context)
     assert "model_path must be str" in str(raise_info.value)
@@ -317,7 +317,7 @@ def test_model_build_03():
 
 def test_model_build_04():
     with pytest.raises(RuntimeError) as raise_info:
-        context = mslite.Context()
+        context = mslite.Context(thread_num=2)
         model = mslite.Model()
         model.build_from_file(model_path="test.ms", model_type=mslite.ModelType.MINDIR_LITE, context=context)
     assert "model_path does not exist" in str(raise_info.value)
@@ -325,7 +325,7 @@ def test_model_build_04():
 
 def get_model():
     cpu_device_info = mslite.CPUDeviceInfo()
-    context = mslite.Context()
+    context = mslite.Context(thread_num=2)
     context.append_device_info(cpu_device_info)
     model = mslite.Model()
     model.build_from_file(model_path="mobilenetv2.ms", model_type=mslite.ModelType.MINDIR_LITE, context=context)

mindspore/lite/test/ut/src/api/model_parallel_runner_test.cc

@@ -129,6 +129,7 @@ TEST_F(ModelParallelRunnerTest, PredictWithoutInput) {
 
   auto context = std::make_shared<Context>();
   ASSERT_NE(nullptr, context);
+  context->SetThreadNum(2);
   auto &device_list = context->MutableDeviceInfo();
   auto device_info = std::make_shared<mindspore::CPUDeviceInfo>();
   ASSERT_NE(nullptr, device_info);

mindspore/lite/test/ut/src/runtime/kernel/arm/cxx_api/model_test.cc

@@ -21,6 +21,9 @@
 #include "include/api/metrics/accuracy.h"
 
 namespace mindspore {
+namespace {
+constexpr int32_t kNumThreads = 2;
+}
 class TestCxxApiLiteModel : public mindspore::CommonTest {
  public:
   TestCxxApiLiteModel() = default;
@@ -40,6 +43,7 @@ TEST_F(TestCxxApiLiteModel, test_build_graph_uninitialized_FAILED) {
   Model model;
   GraphCell graph_cell;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   context->MutableDeviceInfo().push_back(cpu_context);
 
@@ -50,6 +54,7 @@ TEST_F(TestCxxApiLiteModel, test_build_SUCCES) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   context->MutableDeviceInfo().push_back(cpu_context);
 
@@ -66,6 +71,7 @@ TEST_F(TestCxxApiLiteModel, test_train_mode_SUCCES) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   context->MutableDeviceInfo().push_back(cpu_context);
 
@@ -85,6 +91,7 @@ TEST_F(TestCxxApiLiteModel, test_outputs_SUCCESS) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   context->MutableDeviceInfo().push_back(cpu_context);
 
@@ -106,6 +113,7 @@ TEST_F(TestCxxApiLiteModel, test_metrics_SUCCESS) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   context->MutableDeviceInfo().push_back(cpu_context);
 
@@ -121,6 +129,7 @@ TEST_F(TestCxxApiLiteModel, test_getparams_SUCCESS) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   context->MutableDeviceInfo().push_back(cpu_context);
   auto train_cfg = std::make_shared<TrainCfg>();
@@ -157,6 +166,7 @@ TEST_F(TestCxxApiLiteModel, test_getgrads_SUCCESS) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   context->MutableDeviceInfo().push_back(cpu_context);
   auto train_cfg = std::make_shared<TrainCfg>();
@@ -188,6 +198,7 @@ TEST_F(TestCxxApiLiteModel, test_fp32_SUCCESS) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   cpu_context->SetEnableFP16(true);
   context->MutableDeviceInfo().push_back(cpu_context);
@@ -201,6 +212,7 @@ TEST_F(TestCxxApiLiteModel, test_fp16_SUCCESS) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   cpu_context->SetEnableFP16(true);
   context->MutableDeviceInfo().push_back(cpu_context);
@@ -216,6 +228,7 @@ TEST_F(TestCxxApiLiteModel, set_weights_FAILURE) {
   Model model;
   Graph graph;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   cpu_context->SetEnableFP16(true);
   context->MutableDeviceInfo().push_back(cpu_context);
@@ -239,6 +252,7 @@ TEST_F(TestCxxApiLiteModel, set_get_lr_SUCCESS) {
   Graph graph;
   float learn_rate = 0.2;
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
   cpu_context->SetEnableFP16(true);
   context->MutableDeviceInfo().push_back(cpu_context);

tests/ut/cpp/cxx_api/context_test.cc

@@ -18,6 +18,9 @@
 #include "include/api/context.h"
 
 namespace mindspore {
+namespace {
+constexpr int32_t kNumThreads = 2;
+}
 class TestCxxApiContext : public UT::Common {
  public:
   TestCxxApiContext() = default;
@@ -59,6 +62,7 @@ TEST_F(TestCxxApiContext, test_context_get_set_SUCCESS) {
 
 TEST_F(TestCxxApiContext, test_context_cpu_context_SUCCESS) {
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   std::shared_ptr<CPUDeviceInfo> cpu = std::make_shared<CPUDeviceInfo>();
   cpu->SetEnableFP16(true);
   context->MutableDeviceInfo().push_back(cpu);
@@ -81,6 +85,7 @@ TEST_F(TestCxxApiContext, test_context_ascend_context_FAILED) {
   std::string option_9_ans = "1,2,3,4,5";
 
   auto context = std::make_shared<Context>();
+  context->SetThreadNum(kNumThreads);
   std::shared_ptr<AscendDeviceInfo> ascend310 = std::make_shared<AscendDeviceInfo>();
   ascend310->SetInputShape(option_1);
   ascend310->SetInsertOpConfigPath(option_2);