Add TVMC Frontend for PaddlePaddle

python/gen_requirements.py

@@ -146,6 +146,7 @@
                 "future",  # Hidden dependency of torch.
                 "onnx",
                 "onnxruntime",
+                "paddlepaddle",
                 "tensorflow",
                 "tflite",
                 "torch",

python/tvm/driver/tvmc/frontends.py

@@ -265,12 +265,38 @@ def load(self, path, shape_dict=None, **kwargs):
         return relay.frontend.from_pytorch(traced_model, input_shapes, **kwargs)
 
 
+class PaddleFrontend(Frontend):
+    """PaddlePaddle frontend for TVMC"""
+
+    @staticmethod
+    def name():
+        return "paddle"
+
+    @staticmethod
+    def suffixes():
+        return ["pdmodel", "pdiparams"]
+
+    def load(self, path, shape_dict=None, **kwargs):
+        # pylint: disable=C0415
+        import paddle
+
+        paddle.enable_static()
+        paddle.disable_signal_handler()
+
+        # pylint: disable=E1101
+        exe = paddle.static.Executor(paddle.CPUPlace())
+        prog, _, _ = paddle.static.load_inference_model(path, exe)
+
+        return relay.frontend.from_paddle(prog, shape_dict=shape_dict, **kwargs)
+
+
 ALL_FRONTENDS = [
     KerasFrontend,
     OnnxFrontend,
     TensorflowFrontend,
     TFLiteFrontend,
     PyTorchFrontend,
+    PaddleFrontend,
 ]
 
 

tests/python/driver/tvmc/conftest.py

@@ -33,7 +33,7 @@ def download_and_untar(model_url, model_sub_path, temp_dir):
     model_tar_name = os.path.basename(model_url)
     model_path = download_testdata(model_url, model_tar_name, module=["tvmc"])
 
-    if model_path.endswith("tgz") or model_path.endswith("gz"):
+    if model_path.endswith("tgz") or model_path.endswith("gz") or model_path.endswith("tar"):
         tar = tarfile.open(model_path)
         tar.extractall(path=temp_dir)
         tar.close()
@@ -137,6 +137,18 @@ def onnx_resnet50():
     return model_file
 
 
+@pytest.fixture(scope="session")
+def paddle_resnet50(tmpdir_factory):
+    base_url = "https://bj.bcebos.com/x2paddle/models"
+    model_url = "paddle_resnet50.tar"
+    model_file = download_and_untar(
+        "{}/{}".format(base_url, model_url),
+        "paddle_resnet50/model",
+        temp_dir=tmpdir_factory.mktemp("data"),
+    )
+    return model_file
+
+
 @pytest.fixture(scope="session")
 def onnx_mnist():
     base_url = "https://github.com/onnx/models/raw/master/vision/classification/mnist/model"

tests/python/driver/tvmc/test_compiler.py

@@ -273,6 +273,84 @@ def test_cross_compile_options_aarch64_onnx_module(onnx_resnet50):
     assert os.path.exists(dumps_path)
 
 
+def verify_compile_paddle_module(model, shape_dict=None):
+    pytest.importorskip("paddle")
+    tvmc_model = tvmc.load(model, "paddle", shape_dict=shape_dict)
+    tvmc_package = tvmc.compile(tvmc_model, target="llvm", dump_code="ll", desired_layout="NCHW")
+    dumps_path = tvmc_package.package_path + ".ll"
+
+    # check for output types
+    assert type(tvmc_package) is TVMCPackage
+    assert type(tvmc_package.graph) is str
+    assert type(tvmc_package.lib_path) is str
+    assert type(tvmc_package.params) is bytearray
+    assert os.path.exists(dumps_path)
+
+
+def test_compile_paddle_module(paddle_resnet50):
+    # some CI environments wont offer Paddle, so skip in case it is not present
+    pytest.importorskip("paddle")
+    # Check default compilation.
+    verify_compile_paddle_module(paddle_resnet50)
+    # Check with manual shape override
+    shape_string = "inputs:[1,3,224,224]"
+    shape_dict = tvmc.common.parse_shape_string(shape_string)
+    verify_compile_paddle_module(paddle_resnet50, shape_dict)
+
+
+# This test will be skipped if the AArch64 cross-compilation toolchain is not installed.
+@pytest.mark.skipif(
+    not shutil.which("aarch64-linux-gnu-gcc"), reason="cross-compilation toolchain not installed"
+)
+def test_cross_compile_aarch64_paddle_module(paddle_resnet50):
+    # some CI environments wont offer paddle, so skip in case it is not present
+    pytest.importorskip("paddle")
+
+    tvmc_model = tvmc.load(paddle_resnet50, "paddle")
+    tvmc_package = tvmc.compile(
+        tvmc_model,
+        target="llvm -device=arm_cpu -mtriple=aarch64-linux-gnu -mattr=+neon",
+        dump_code="asm",
+        cross="aarch64-linux-gnu-gcc",
+    )
+    dumps_path = tvmc_package.package_path + ".asm"
+
+    # check for output types
+    assert type(tvmc_package) is TVMCPackage
+    assert type(tvmc_package.graph) is str
+    assert type(tvmc_package.lib_path) is str
+    assert type(tvmc_package.params) is bytearray
+    assert os.path.exists(dumps_path)
+
+
+# This test will be skipped if the AArch64 cross-compilation toolchain is not installed.
+@pytest.mark.skipif(
+    not shutil.which("aarch64-linux-gnu-gcc"), reason="cross-compilation toolchain not installed"
+)
+def test_cross_compile_options_aarch64_paddle_module(paddle_resnet50):
+    # some CI environments wont offer paddle, so skip in case it is not present
+    pytest.importorskip("paddle")
+
+    fake_sysroot_dir = utils.tempdir().relpath("")
+
+    tvmc_model = tvmc.load(paddle_resnet50, "paddle")
+    tvmc_package = tvmc.compile(
+        tvmc_model,
+        target="llvm -device=arm_cpu -mtriple=aarch64-linux-gnu -mattr=+neon",
+        dump_code="asm",
+        cross="aarch64-linux-gnu-gcc",
+        cross_options="--sysroot=" + fake_sysroot_dir,
+    )
+    dumps_path = tvmc_package.package_path + ".asm"
+
+    # check for output types
+    assert type(tvmc_package) is TVMCPackage
+    assert type(tvmc_package.graph) is str
+    assert type(tvmc_package.lib_path) is str
+    assert type(tvmc_package.params) is bytearray
+    assert os.path.exists(dumps_path)
+
+
 @tvm.testing.requires_opencl
 def test_compile_opencl(tflite_mobilenet_v1_0_25_128):
     pytest.importorskip("tflite")

tests/python/driver/tvmc/test_frontends.py

@@ -84,6 +84,14 @@ def test_guess_frontend_tensorflow():
     assert type(sut) is tvmc.frontends.TensorflowFrontend
 
 
+def test_guess_frontend_paddle():
+    # some CI environments wont offer Paddle, so skip in case it is not present
+    pytest.importorskip("paddle")
+
+    sut = tvmc.frontends.guess_frontend("a_model.pdmodel")
+    assert type(sut) is tvmc.frontends.PaddleFrontend
+
+
 def test_guess_frontend_invalid():
     with pytest.raises(TVMCException):
         tvmc.frontends.guess_frontend("not/a/file.txt")
@@ -161,6 +169,16 @@ def test_load_model__pb(pb_mobilenet_v1_1_quant):
     assert "MobilenetV1/Conv2d_0/weights" in tvmc_model.params.keys()
 
 
+def test_load_model__paddle(paddle_resnet50):
+    # some CI environments wont offer Paddle, so skip in case it is not present
+    pytest.importorskip("paddle")
+
+    tvmc_model = tvmc.load(paddle_resnet50, model_format="paddle")
+    assert type(tvmc_model) is TVMCModel
+    assert type(tvmc_model.mod) is IRModule
+    assert type(tvmc_model.params) is dict
+
+
 def test_load_model___wrong_language__to_keras(tflite_mobilenet_v1_1_quant):
     # some CI environments wont offer TensorFlow/Keras, so skip in case it is not present
     pytest.importorskip("tensorflow")

tests/python/driver/tvmc/test_tvmc_common.py

@@ -75,6 +75,31 @@ def _is_layout_transform(node):
     assert any(layout_transform_calls), "Expected 'layout_transform NCWH->NHWC' not found"
 
 
+def test_compile_paddle_module_nchw_to_nhwc(paddle_resnet50):
+    # some CI environments wont offer Paddle, so skip in case it is not present
+    pytest.importorskip("paddle")
+
+    tvmc_model = tvmc.frontends.load_model(paddle_resnet50, "paddle")
+    before = tvmc_model.mod
+
+    expected_layout = "NHWC"
+    after = tvmc.common.convert_graph_layout(before, expected_layout)
+
+    layout_transform_calls = []
+
+    def _is_layout_transform(node):
+        if isinstance(node, tvm.relay.expr.Call):
+            layout_transform_calls.append(
+                node.op.name == "layout_transform"
+                and node.attrs.src_layout == "NCHW"
+                and node.attrs.dst_layout == "NHWC"
+            )
+
+    tvm.relay.analysis.post_order_visit(after["main"], _is_layout_transform)
+
+    assert any(layout_transform_calls), "Expected 'layout_transform NCWH->NHWC' not found"
+
+
 def test_compile_tflite_module__same_layout__nhwc_to_nhwc(tflite_mobilenet_v1_1_quant):
     # some CI environments wont offer TFLite, so skip in case it is not present
     pytest.importorskip("tflite")