Use the new bucket's idx in call to sampleselect_filter for cuda/hip and dpcpp

Author: nbeams

common/cuda_hip/factorization/par_ilut_select_kernels.cpp

@@ -119,7 +119,7 @@ void threshold_select(std::shared_ptr<const DefaultExecutor> exec,
                            partial_counts, total_counts);
         auto new_bucket = sampleselect_find_bucket(exec, total_counts, rank);
         sampleselect_filter(exec, tmp_in, bucket.size, oracles, partial_counts,
-                            bucket.idx, tmp_out);
+                            new_bucket.idx, tmp_out);
 
         rank -= new_bucket.begin;
         bucket.size = new_bucket.size;

dpcpp/factorization/par_ilut_select_kernel.dp.cpp

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: 2017 - 2024 The Ginkgo authors
+// SPDX-FileCopyrightText: 2017 - 2025 The Ginkgo authors
 //
 // SPDX-License-Identifier: BSD-3-Clause
 
@@ -122,7 +122,7 @@ void threshold_select(std::shared_ptr<const DefaultExecutor> exec,
                            partial_counts, total_counts);
         auto new_bucket = sampleselect_find_bucket(exec, total_counts, rank);
         sampleselect_filter(exec, tmp_in, bucket.size, oracles, partial_counts,
-                            bucket.idx, tmp_out);
+                            new_bucket.idx, tmp_out);
 
         rank -= new_bucket.begin;
         bucket.size = new_bucket.size;