add npu adaptation for assign_score_withk op (#3209)

* test

* fix

* lint fix

* lint1

mmcv/ops/csrc/pytorch/npu/assign_score_withk_npu.cpp

@@ -0,0 +1,23 @@
+#include "pytorch_npu_helper.hpp"
+
+using namespace NPU_NAME_SPACE;
+using namespace std;
+
+void assign_score_withk_forward_npu(int B, int N0, int N1, int M, int K, int O,
+                                    int aggregate, const Tensor& points,
+                                    const Tensor& centers, const Tensor& scores,
+                                    const Tensor& knn_idx, Tensor& output) {
+  at::Tensor points_trans = points.permute({0, 3, 1, 2});
+  at::Tensor centers_trans = centers.permute({0, 3, 1, 2});
+  EXEC_NPU_CMD(aclnnAssignScoreWithk, points_trans, centers_trans, scores,
+               knn_idx, B, N0, N1, M, K, O, aggregate, output);
+}
+
+void assign_score_withk_forward_impl(int B, int N0, int N1, int M, int K, int O,
+                                     int aggregate, const Tensor& points,
+                                     const Tensor& centers,
+                                     const Tensor& scores,
+                                     const Tensor& knn_idx, Tensor& output);
+
+REGISTER_NPU_IMPL(assign_score_withk_forward_impl,
+                  assign_score_withk_forward_npu);