Update onnx app to Adams2019 autoscheduler and new autoscheduler API (h…

…alide#7673)

* Update onnx app to Adams2019 autoscheduler and new autoscheduler API

Fixes halide#7670

* Add model test too

* Remove use of tmpnam

* Don't test onnx app in a 32-bit build

apps/onnx/Makefile

@@ -1,5 +1,11 @@
 include ../support/Makefile.inc
 
+ifneq (,$(findstring -m32,$(CXX) $(CC) $(CCFLAGS) $(CXXFLAGS)))
+build:
+	@echo "Not testing this app in a 32-bit build (-m32 found in flags)"
+test: build
+else
+
 PROTOC := $(shell which protoc)
 
 ifdef PROTOC
@@ -72,7 +78,7 @@ $(BIN)/%/onnx_converter_generator_test: onnx_converter_generator_test.cc $(BIN)/
 
 build: $(BIN)/$(HL_TARGET)/onnx_converter_test $(BIN)/$(HL_TARGET)/onnx_converter_generator_test
 
-test: build
+test: build model_test
 	LD_LIBRARY_PATH=$(BIN) $(BIN)/$(HL_TARGET)/onnx_converter_test
 	LD_LIBRARY_PATH=$(BIN) $(BIN)/$(HL_TARGET)/onnx_converter_generator_test
 
@@ -88,7 +94,7 @@ PYCXXFLAGS =  $(CXXFLAGS) $(PYBIND11_CFLAGS) -Wno-deprecated-register
 
 # Python extension for HalideModel
 $(BIN)/%/$(PY_MODEL_EXT): model.cpp $(BIN)/%/oclib.a
-	$(CXX) $(PYCXXFLAGS) -Wall -shared -fPIC -I$(BIN)/$* $^ $(LIBHALIDE_LDFLAGS) -o $@ $(LDFLAGS)
+	$(CXX) $(PYCXXFLAGS) -Wall -shared -fPIC -I$(BIN)/$* $^ $(LIBHALIDE_LDFLAGS) -Wl,--no-as-needed -lautoschedule_adams2019 -Wl,--as-needed -o $@ $(LDFLAGS)
 
 
 model_test: $(BIN)/$(HL_TARGET)/$(PY_MODEL_EXT)
@@ -104,3 +110,4 @@ build:
 test: build
 
 endif
+endif

apps/onnx/model.cpp

@@ -65,7 +65,8 @@ HalideModel convert_onnx_model(
 std::string auto_schedule(const HalideModel &pipeline) {
     // Generate a schedule for the pipeline.
     Halide::Target tgt = Halide::get_host_target();
-    auto schedule = pipeline.rep->auto_schedule(tgt);
+    Halide::AutoschedulerParams autoscheduler_params = Halide::AutoschedulerParams("Adams2019");
+    auto schedule = pipeline.rep->apply_autoscheduler(tgt, autoscheduler_params);
     return schedule.schedule_source;
 }
 
@@ -528,15 +529,15 @@ void print_loop_nest(const HalideModel &pipeline) {
 }
 
 void print_lowered_statement(const HalideModel &pipeline) {
-    std::string tmp_file = std::tmpnam(nullptr);
+    Halide::Internal::TemporaryFile f("model", ".stmt");
+    std::string tmp_file = f.pathname();
     pipeline.rep->compile_to_lowered_stmt(
         tmp_file, pipeline.rep->infer_arguments());
     std::ifstream is(tmp_file);
     std::string line;
     while (std::getline(is, line)) {
         std::cout << line << "\n";
     }
-    std::remove(tmp_file.c_str());
 }
 
 }  // namespace

apps/onnx/model_test.py

@@ -35,10 +35,10 @@ def test_small_model(self):
         model.BuildFromOnnxModel(onnx_model)
         schedule = model.OptimizeSchedule()
         schedule = schedule.replace('\n', ' ')
-        expected_schedule = r'// --- BEGIN machine-generated schedule // Target: .+// MachineParams: .+// Delete this line if not using Generator Pipeline pipeline = get_pipeline\(\);.+Func OUT = pipeline.get_func\(1\);.+{.+}.+'
+        expected_schedule = r'.*Func OUT = pipeline.get_func\(1\);.+'
         self.assertRegex(schedule, expected_schedule)
 
-        input = np.random.rand(2, 3) - 0.5
+        input = (np.random.rand(2, 3) - 0.5).astype('float32')
         outputs = model.run([input])
         self.assertEqual(1, len(outputs))
         output = outputs[0]
@@ -62,12 +62,12 @@ def test_scalars(self):
         model = Model()
         model.BuildFromOnnxModel(onnx_model)
         schedule = model.OptimizeSchedule()
-        schedule = schedule.replace('\n', ' ')
-        expected_schedule = r'// --- BEGIN machine-generated schedule // Target: .+// MachineParams: .+// Delete this line if not using Generator Pipeline pipeline = get_pipeline\(\);.+Func C = pipeline.get_func\(2\);.+{.+}.+'
+        schedule = schedule.replace('\n', ' ')        
+        expected_schedule = r'.*Func C = pipeline.get_func\(2\);.+'
         self.assertRegex(schedule, expected_schedule)
 
-        input1 = np.random.randint(-10, 10, size=())
-        input2 = np.random.randint(-10, 10, size=())
+        input1 = np.random.randint(-10, 10, size=()).astype('int32')
+        input2 = np.random.randint(-10, 10, size=()).astype('int32')
         outputs = model.run([input1, input2])
         self.assertEqual(1, len(outputs))
         output = outputs[0]