Merge pull request #17848 from rapidsai/branch-25.02

Forward-merge branch-25.02 into branch-25.04
rapidsai · Jan 29, 2025 · 9817d01 · 9817d01
2 parents 9f4afb4 + 95c69c3
commit 9817d01
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 6 deletions.
diff --git a/cpp/src/io/parquet/decode_fixed.cu b/cpp/src/io/parquet/decode_fixed.cu
@@ -1167,8 +1167,10 @@ CUDF_KERNEL void __launch_bounds__(decode_block_size_t, 8)
     // For large strings, update the initial string buffer offset to be used during large string
     // column construction. Otherwise, convert string sizes to final offsets.
     if (s->col.is_large_string_col) {
-      compute_initial_large_strings_offset(
-        s, initial_str_offsets[pages[page_idx].chunk_idx], has_lists_t);
+      // page.chunk_idx are ordered by input_col_idx and row_group_idx respectively.
+      auto const chunks_per_rowgroup = initial_str_offsets.size();
+      auto const input_col_idx       = pages[page_idx].chunk_idx % chunks_per_rowgroup;
+      compute_initial_large_strings_offset(s, initial_str_offsets[input_col_idx], has_lists_t);
     } else {
       convert_small_string_lengths_to_offsets<decode_block_size_t>(s, has_lists_t);
     }

diff --git a/cpp/src/io/parquet/page_delta_decode.cu b/cpp/src/io/parquet/page_delta_decode.cu
@@ -583,8 +583,10 @@ CUDF_KERNEL void __launch_bounds__(decode_block_size)
   // For large strings, update the initial string buffer offset to be used during large string
   // column construction. Otherwise, convert string sizes to final offsets.
   if (s->col.is_large_string_col) {
-    compute_initial_large_strings_offset(
-      s, initial_str_offsets[pages[page_idx].chunk_idx], has_repetition);
+    // page.chunk_idx are ordered by input_col_idx and row_group_idx respectively.
+    auto const chunks_per_rowgroup = initial_str_offsets.size();
+    auto const input_col_idx       = pages[page_idx].chunk_idx % chunks_per_rowgroup;
+    compute_initial_large_strings_offset(s, initial_str_offsets[input_col_idx], has_repetition);
   } else {
     convert_small_string_lengths_to_offsets<decode_block_size>(s, has_repetition);
   }
@@ -742,8 +744,10 @@ CUDF_KERNEL void __launch_bounds__(decode_block_size)
   // For large strings, update the initial string buffer offset to be used during large string
   // column construction. Otherwise, convert string sizes to final offsets.
   if (s->col.is_large_string_col) {
-    compute_initial_large_strings_offset(
-      s, initial_str_offsets[pages[page_idx].chunk_idx], has_repetition);
+    // page.chunk_idx are ordered by input_col_idx and row_group_idx respectively.
+    auto const chunks_per_rowgroup = initial_str_offsets.size();
+    auto const input_col_idx       = pages[page_idx].chunk_idx % chunks_per_rowgroup;
+    compute_initial_large_strings_offset(s, initial_str_offsets[input_col_idx], has_repetition);
   } else {
     convert_small_string_lengths_to_offsets<decode_block_size>(s, has_repetition);
   }