In-place guarantees for RLE

NVIDIA · May 31, 2022 · d4ac304 · d4ac304
1 parent 4382cfb
commit d4ac304
Showing 1 changed file with 10 additions and 0 deletions.
diff --git a/cub/device/device_run_length_encode.cuh b/cub/device/device_run_length_encode.cuh
@@ -85,6 +85,11 @@ struct DeviceRunLengthEncode
    * - The total number of runs encountered is written to `d_num_runs_out`.
    * - The `==` equality operator is used to determine whether values are 
    *   equivalent
+   * - Let `out` represent a set of following ranges 
+   *   `[d_unique_out, d_unique_out + *d_num_runs_out)`,
+   *   `[d_counts_out, d_counts_out + *d_num_runs_out)`,
+   *   `d_num_runs_out`. There shall be no overlap between ranges represented
+   *   by `out`. The range `[d_in, d_in + num_items)` shall not overlap `out`.
    * - @devicestorage
    *
    * @par Performance
@@ -249,6 +254,11 @@ struct DeviceRunLengthEncode
    * - The total number of runs encountered is written to `d_num_runs_out`.
    * - The `==` equality operator is used to determine whether values are 
    *   equivalent
+   * - Let `out` represent a set of following ranges 
+   *   `[d_offsets_out, d_offsets_out + *d_num_runs_out)`,
+   *   `[d_lengths_out, d_lengths_out + *d_num_runs_out)`,
+   *   `d_num_runs_out`. There shall be no overlap between ranges represented
+   *   by `out`. The range `[d_in, d_in + num_items)` shall not overlap `out`.
    * - @devicestorage
    *
    * @par Performance