[Bug] Fix ndarray memory leak

gstaichi/program/program.h

@@ -302,6 +302,10 @@ class TI_DLL_EXPORT Program {
     return program_impl_.get();
   }
 
+  size_t unsupported_get_num_ndarrays() const {
+    return ndarrays_.size();
+  }
+
   // TODO(zhanlue): Move these members and corresponding interfaces to
   // ProgramImpl Ideally, Program should serve as a pure interface class and all
   // the implementations should fall inside ProgramImpl

gstaichi/python/export_lang.cpp

@@ -370,6 +370,8 @@ void export_lang(py::module &m) {
            [](Program *program, SNode *snode, int element_ndim, int n, int m) {
              return field_to_dlpack(program, snode, element_ndim, n, m);
            })
+      .def("_unsupported_get_num_ndarrays",
+           &Program::unsupported_get_num_ndarrays)
       .def("config", &Program::compile_config,
            py::return_value_policy::reference)
       .def("sync_kernel_profiler",

python/gstaichi/lang/_ndarray.py

@@ -38,6 +38,14 @@ def __init__(self):
         # we register with runtime, in order to enable reset to work later
         impl.get_runtime().ndarrays.add(self)
 
+    def __del__(self):
+        if impl is not None and impl.get_runtime is not None and impl.get_runtime() is not None:
+            arr = getattr(self, "arr")
+            if arr is not None:
+                prog = impl.get_runtime()._prog
+                if prog is not None:
+                    prog.delete_ndarray(self.arr)
+
     def to_dlpack(self):
         return impl.get_runtime().prog.ndarray_to_dlpack(self, self.arr)
 
@@ -270,12 +278,6 @@ def __init__(self, dtype, arr_shape):
         self.shape = tuple(self.arr.shape)
         self.element_type = dtype
 
-    def __del__(self):
-        if impl is not None and impl.get_runtime is not None and impl.get_runtime() is not None:
-            prog = impl.get_runtime()._prog
-            if prog is not None:
-                prog.delete_ndarray(self.arr)
-
     @property
     def element_shape(self):
         return ()

tests/python/test_ndarray.py

@@ -1199,3 +1199,14 @@ def my_kernel({args}) -> None:
         my_kernel(*arg_objs_l)
     for i in range(num_args):
         assert arg_objs_l[i][0] == i + 1
+
+
+@pytest.mark.parametrize("dtype", [ti.i32, ti.types.vector(3, ti.f32), ti.types.matrix(2, 2, ti.f32)])
+@test_utils.test()
+def test_ndarray_del(dtype) -> None:
+    def foo():
+        nd = ti.ndarray(dtype, (1000,))
+        assert ti.lang.impl.get_runtime().prog._unsupported_get_num_ndarrays() == 1
+
+    foo()
+    assert ti.lang.impl.get_runtime().prog._unsupported_get_num_ndarrays() == 0