[CustomDevice] add eager mode support (#42034)

paddle/fluid/pybind/eager.cc

@@ -146,10 +146,13 @@ void InitTensorWithNumpyValue(TensorObject* self, const py::object& array,
                                                     zero_copy);
   } else if (platform::is_npu_place(place)) {
     SetTensorFromPyArray<platform::NPUPlace>(impl_ptr, array, place, zero_copy);
+  } else if (platform::is_custom_place(place)) {
+    SetTensorFromPyArray<platform::CustomPlace>(impl_ptr, array, place,
+                                                zero_copy);
   } else {
     PADDLE_THROW(platform::errors::InvalidArgument(
         "Place should be one of "
-        "CPUPlace/XPUPlace/CUDAPlace/CUDAPinnedPlace/NPUPlace"));
+        "CPUPlace/XPUPlace/CUDAPlace/CUDAPinnedPlace/NPUPlace/CustomPlace"));
   }
 }
 

paddle/fluid/pybind/eager_utils.cc

@@ -46,6 +46,7 @@ extern PyTypeObject* g_cpuplace_pytype;
 extern PyTypeObject* g_xpuplace_pytype;
 extern PyTypeObject* g_npuplace_pytype;
 extern PyTypeObject* g_cudapinnedplace_pytype;
+extern PyTypeObject* g_customplace_pytype;
 extern PyTypeObject* g_framework_tensor_pytype;
 extern PyTypeObject* g_framework_lodtensorarray_pytype;
 extern PyTypeObject* g_custom_op_kernel_ctx_pytype;
@@ -377,10 +378,15 @@ platform::Place CastPyArg2Place(PyObject* obj, ssize_t arg_pos) {
   } else if (PyObject_IsInstance(
                  obj, reinterpret_cast<PyObject*>(g_cudapinnedplace_pytype))) {
     place = ::pybind11::handle(obj).cast<platform::CUDAPinnedPlace>();
+  } else if (PyObject_IsInstance(
+                 obj, reinterpret_cast<PyObject*>(g_customplace_pytype))) {
+    place = ::pybind11::handle(obj).cast<platform::CustomPlace>();
   } else {
     PADDLE_THROW(platform::errors::InvalidArgument(
         "argument (position %d) must be "
-        "one of(Place,CUDAPlace,CPUPlace,XPUPlace,NPUPlace,CUDAPinnedPlace), "
+        "one "
+        "of(Place,CUDAPlace,CPUPlace,XPUPlace,NPUPlace,CUDAPinnedPlace,"
+        "CustomPlace), "
         "but got %s",
         arg_pos + 1, reinterpret_cast<PyTypeObject*>(obj->ob_type)->tp_name));
   }

paddle/fluid/pybind/pybind.cc

@@ -193,6 +193,7 @@ PyTypeObject *g_xpuplace_pytype = nullptr;
 PyTypeObject *g_npuplace_pytype = nullptr;
 PyTypeObject *g_cudapinnedplace_pytype = nullptr;
 PyTypeObject *g_mluplace_pytype = nullptr;
+PyTypeObject *g_customplace_pytype = nullptr;
 PyTypeObject *g_framework_tensor_pytype = nullptr;
 PyTypeObject *g_framework_lodtensorarray_pytype = nullptr;
 PyTypeObject *g_custom_op_kernel_ctx_pytype = nullptr;
@@ -2125,8 +2126,8 @@ All parameter, weight, gradient are variables in Paddle.
 #endif
     return devices;
   });
-  py::class_<platform::CustomPlace>(m, "CustomPlace",
-                                    R"DOC(
+  py::class_<platform::CustomPlace> customplace(m, "CustomPlace",
+                                                R"DOC(
     CustomPlace is a descriptor of a device.
     It represents a custom device on which a tensor will be allocated and a model will run.
 
@@ -2135,7 +2136,9 @@ All parameter, weight, gradient are variables in Paddle.
 
           import paddle
           fake_cpu_place = paddle.CustomPlace("FakeCPU", 0)
-                                             )DOC")
+                                             )DOC");
+  g_customplace_pytype = reinterpret_cast<PyTypeObject *>(customplace.ptr());
+  customplace
       .def("__init__",
            [](platform::CustomPlace &self, const std::string &device_type,
               int dev_id) {