[Bugfix] Fix bugs of farthest_point_sampler (dmlc#3327)

* fix start_idx * fix the bug when cuda > 0 Co-authored-by: Tong He <[email protected]>
standbyme · Sep 10, 2021 · 6454c79 · 6454c79
1 parent d17b299
commit 6454c79
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 2 deletions.
diff --git a/python/dgl/geometry/fps.py b/python/dgl/geometry/fps.py
@@ -8,6 +8,7 @@
 
 __all__ = ['farthest_point_sampler']
 
+
 def farthest_point_sampler(pos, npoints, start_idx=None):
     """Farthest Point Sampler without the need to compute all pairs of distance.
 
@@ -49,12 +50,13 @@ def farthest_point_sampler(pos, npoints, start_idx=None):
     pos = pos.reshape(-1, C)
     dist = F.zeros((B * N), dtype=pos.dtype, ctx=ctx)
     if start_idx is None:
-        start_idx = F.randint(shape=(B, ), dtype=F.int64, ctx=ctx, low=0, high=N-1)
+        start_idx = F.randint(shape=(B, ), dtype=F.int64,
+                              ctx=ctx, low=0, high=N-1)
     else:
         if start_idx >= N or start_idx < 0:
             raise DGLError("Invalid start_idx, expected 0 <= start_idx < {}, got {}".format(
                 N, start_idx))
-        start_idx = F.full_1d((B, ), start_idx, dtype=F.int64, ctx=ctx)
+        start_idx = F.full_1d(B, start_idx, dtype=F.int64, ctx=ctx)
     result = F.zeros((npoints * B), dtype=F.int64, ctx=ctx)
     _farthest_point_sampler(pos, B, npoints, dist, start_idx, result)
     return result.reshape(B, npoints)
diff --git a/src/geometry/cuda/geometry_op_impl.cu b/src/geometry/cuda/geometry_op_impl.cu
@@ -107,6 +107,7 @@ void FarthestPointSampler(NDArray array, int64_t batch_size, int64_t sample_poin
 
   // sample for each cloud in the batch
   IdType* start_idx_data = static_cast<IdType*>(start_idx->data);
+  CUDA_CALL(cudaSetDevice(array->ctx.device_id));
 
   CUDA_KERNEL_CALL(fps_kernel,
       batch_size, THREADS, 0, thr_entry->stream,

diff --git a/tests/pytorch/test_geometry.py b/tests/pytorch/test_geometry.py
@@ -25,6 +25,18 @@ def test_fps():
     assert res.sum() > 0
 
 
+def test_fps_start_idx():
+    N = 1000
+    batch_size = 5
+    sample_points = 10
+    x = th.tensor(np.random.uniform(size=(batch_size, int(N/batch_size), 3)))
+    ctx = F.ctx()
+    if F.gpu_ctx():
+        x = x.to(ctx)
+    res = farthest_point_sampler(x, sample_points, start_idx=0)
+    assert th.any(res[:, 0] == 0)
+
+
 @pytest.mark.parametrize('algorithm', ['bruteforce-blas', 'bruteforce', 'kd-tree'])
 @pytest.mark.parametrize('dist', ['euclidean', 'cosine'])
 def test_knn_cpu(algorithm, dist):
@@ -208,4 +220,5 @@ def test_edge_coarsening(idtype, g, weight, relabel):
 
 if __name__ == '__main__':
     test_fps()
+    test_fps_start_idx()
     test_knn()