icexelloss commented on code in PR #13880:
URL: https://github.com/apache/arrow/pull/13880#discussion_r954255104
##########
cpp/src/arrow/compute/exec/asof_join_node.cc:
##########
@@ -137,18 +173,88 @@ struct MemoStore {
}
};
+// a specialized higher-performance variation of Hashing64 logic from
hash_join_node
+// the code here avoids recreating objects that are independent of each batch
processed
+class KeyHasher {
+ static constexpr int kMiniBatchLength = util::MiniBatch::kMiniBatchLength;
+
+ public:
+ explicit KeyHasher(const vec_col_index_t& indices)
+ : indices_(indices),
+ metadata_(indices.size()),
+ batch_(NULLPTR),
+ hashes_(),
+ ctx_(),
+ column_arrays_(),
+ stack_() {
+ ctx_.stack = &stack_;
+ column_arrays_.resize(indices.size());
+ }
+
+ Status Init(ExecContext* exec_context, const std::shared_ptr<arrow::Schema>&
schema) {
+ ctx_.hardware_flags = exec_context->cpu_info()->hardware_flags();
+ const auto& fields = schema->fields();
+ for (size_t k = 0; k < metadata_.size(); k++) {
+ ARROW_ASSIGN_OR_RAISE(metadata_[k],
+
ColumnMetadataFromDataType(fields[indices_[k]]->type()));
+ }
+ return stack_.Init(exec_context->memory_pool(),
+ 4 * kMiniBatchLength * sizeof(uint32_t));
+ }
+
+ const std::vector<HashType>& HashesFor(const RecordBatch* batch) {
+ if (batch_ == batch) {
+ return hashes_;
Review Comment:
Why do we need to cache the `hashes_` here? (i.e., why do we call this
method twice for the same batch)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscr...@arrow.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
