[Enchancement](sort) some optimization for full sort (#45824)

### What problem does this PR solve?

![QQ_1734963592703](https://github.com/user-attachments/assets/d5a93341-d000-4475-8435-792b521a8d00)

### Release note

None

### Check List (For Author)

- Test <!-- At least one of them must be included. -->
    - [ ] Regression test
    - [ ] Unit Test
    - [ ] Manual test (add detailed scripts or steps below)
    - [ ] No need to test or manual test. Explain why:
- [ ] This is a refactor/code format and no logic has been changed.
        - [ ] Previous test can cover this change.
        - [ ] No code files have been changed.
        - [ ] Other reason <!-- Add your reason?  -->

- Behavior changed:
    - [ ] No.
    - [ ] Yes. <!-- Explain the behavior change -->

- Does this need documentation?
    - [ ] No.
- [ ] Yes. <!-- Add document PR link here. eg:
apache/doris-website#1214 -->

### Check List (For Reviewer who merge this PR)

- [ ] Confirm the release note
- [ ] Confirm test cases
- [ ] Confirm document
- [ ] Add branch pick label <!-- Add branch pick label that this PR
should merge into -->

be/src/vec/common/sort/sorter.cpp

@@ -108,10 +108,6 @@ Status MergeSorterState::merge_sort_read(doris::vectorized::Block* block, int ba
 
 Status MergeSorterState::_merge_sort_read_impl(int batch_size, doris::vectorized::Block* block,
                                                bool* eos) {
-    if (priority_queue_.empty()) {
-        *eos = true;
-        return Status::OK();
-    }
     size_t num_columns = priority_queue_.top().impl->block->columns();
 
     MutableBlock m_block = VectorizedUtils::build_mutable_mem_reuse_block(
@@ -120,13 +116,15 @@ Status MergeSorterState::_merge_sort_read_impl(int batch_size, doris::vectorized
 
     /// Take rows from queue in right order and push to 'merged'.
     size_t merged_rows = 0;
-    while (!priority_queue_.empty()) {
+    // process single element queue on merge_sort_read()
+    while (priority_queue_.size() > 1 && merged_rows < batch_size) {
         auto current = priority_queue_.top();
         priority_queue_.pop();
 
         if (offset_ == 0) {
-            for (size_t i = 0; i < num_columns; ++i)
+            for (size_t i = 0; i < num_columns; ++i) {
                 merged_columns[i]->insert_from(*current->block->get_columns()[i], current->pos);
+            }
             ++merged_rows;
         } else {
             offset_--;
@@ -136,18 +134,9 @@ Status MergeSorterState::_merge_sort_read_impl(int batch_size, doris::vectorized
             current->next();
             priority_queue_.push(current);
         }
-
-        if (merged_rows == batch_size) {
-            break;
-        }
-    }
-    block->set_columns(std::move(merged_columns));
-
-    if (merged_rows == 0) {
-        *eos = true;
-        return Status::OK();
     }
 
+    block->set_columns(std::move(merged_columns));
     return Status::OK();
 }
 
@@ -217,9 +206,15 @@ FullSorter::FullSorter(VSortExecExprs& vsort_exec_exprs, int limit, int64_t offs
 
 Status FullSorter::append_block(Block* block) {
     DCHECK(block->rows() > 0);
+
+    if (_reach_limit() && block->bytes() > _state->unsorted_block_->allocated_bytes() -
+                                                   _state->unsorted_block_->bytes()) {
+        RETURN_IF_ERROR(_do_sort());
+    }
+
     {
         SCOPED_TIMER(_merge_block_timer);
-        auto& data = _state->unsorted_block_->get_columns_with_type_and_name();
+        const auto& data = _state->unsorted_block_->get_columns_with_type_and_name();
         const auto& arrival_data = block->get_columns_with_type_and_name();
         auto sz = block->rows();
         for (int i = 0; i < data.size(); ++i) {
@@ -232,9 +227,6 @@ Status FullSorter::append_block(Block* block) {
         }
         block->clear_column_data();
     }
-    if (_reach_limit()) {
-        RETURN_IF_ERROR(_do_sort());
-    }
     return Status::OK();
 }
 

be/src/vec/common/sort/sorter.h

@@ -177,21 +177,15 @@ class FullSorter final : public Sorter {
 
 private:
     bool _reach_limit() {
-        return _state->unsorted_block_->rows() > buffered_block_size_ ||
-               _state->unsorted_block_->bytes() > buffered_block_bytes_;
+        return _state->unsorted_block_->allocated_bytes() >= buffered_block_bytes_;
     }
 
     Status _do_sort();
 
     std::unique_ptr<MergeSorterState> _state;
 
-    static constexpr size_t INITIAL_BUFFERED_BLOCK_SIZE = 1024 * 1024;
-    static constexpr size_t INITIAL_BUFFERED_BLOCK_BYTES = 64 << 20;
+    static constexpr size_t INITIAL_BUFFERED_BLOCK_BYTES = 64 * 1024 * 1024;
 
-    static constexpr size_t SPILL_BUFFERED_BLOCK_SIZE = 4 * 1024 * 1024;
-    static constexpr size_t SPILL_BUFFERED_BLOCK_BYTES = 256 << 20;
-
-    size_t buffered_block_size_ = INITIAL_BUFFERED_BLOCK_SIZE;
     size_t buffered_block_bytes_ = INITIAL_BUFFERED_BLOCK_BYTES;
 };