diff --git a/src/rapids_singlecell/preprocessing/_kernels/_bbknn.py b/src/rapids_singlecell/preprocessing/_kernels/_bbknn.py
index d6bb2b07..e8607efc 100644
--- a/src/rapids_singlecell/preprocessing/_kernels/_bbknn.py
+++ b/src/rapids_singlecell/preprocessing/_kernels/_bbknn.py
@@ -64,6 +64,7 @@
 extern "C" __global__
 void cut_smaller(
     const int *indptr,
+    const int * index,
     float *data,
     float* vals,
     int n_rows) {
@@ -74,13 +75,15 @@
     int start_idx = indptr[row_id];
     int stop_idx = indptr[row_id+1];
 
-    float cut = vals[row_id];
+    float cut_row = vals[row_id];
     for(int i = start_idx+threadIdx.x; i < stop_idx; i+= blockDim.x){
+        float cut = max(vals[index[i]], cut_row);
         if(data[i]<cut){
             data[i] = 0;
         }
 
     }
+
 }
 """
 cut_smaller_func = cp.RawKernel(_cut_smaller_kernel, "cut_smaller")
diff --git a/src/rapids_singlecell/preprocessing/_neighbors.py b/src/rapids_singlecell/preprocessing/_neighbors.py
index 7aea8c86..db17f2dc 100644
--- a/src/rapids_singlecell/preprocessing/_neighbors.py
+++ b/src/rapids_singlecell/preprocessing/_neighbors.py
@@ -280,15 +280,12 @@ def _trimming(cnts: cp_sparse.csr_matrix, trim: int) -> cp_sparse.csr_matrix:
         (cnts.data, cnts.indptr, cnts.shape[0], trim, vals_gpu),
         shared_mem=shared_mem_size,
     )
-
-    for _ in range(2):
-        cut_smaller_func(
-            (cnts.shape[0],),
-            (64,),
-            (cnts.indptr, cnts.data, vals_gpu, cnts.shape[0]),
-        )
-        cnts.eliminate_zeros()
-        cnts = cnts.T.tocsr()
+    cut_smaller_func(
+        (cnts.shape[0],),
+        (64,),
+        (cnts.indptr, cnts.indices, cnts.data, vals_gpu, cnts.shape[0]),
+    )
+    cnts.eliminate_zeros()
     return cnts