[NSE-1071] Add tiny optimizations for hash aggregation functions (#1072)

* Optimize code logic by considering input null count * Pass input list in const reference * Remove pieces of unnecessary code
oap-project · Aug 21, 2022 · 49d5985 · 49d5985
1 parent 25fa211
commit 49d5985
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 40 deletions.
diff --git a/native-sql-engine/cpp/src/codegen/arrow_compute/ext/actions_impl.cc b/native-sql-engine/cpp/src/codegen/arrow_compute/ext/actions_impl.cc
@@ -50,14 +50,14 @@ struct FindAccumulatorType<I, arrow::enable_if_floating_point<I>> {
   using Type = arrow::DoubleType;
 };
 
-arrow::Status ActionBase::Submit(ArrayList in, int max_group_id,
+arrow::Status ActionBase::Submit(const ArrayList& in, int max_group_id,
                                  std::function<arrow::Status(int)>* on_valid,
                                  std::function<arrow::Status()>* on_null) {
   return arrow::Status::NotImplemented("ActionBase Submit is abstract.");
 }
 
 arrow::Status ActionBase::Submit(
-    std::vector<std::shared_ptr<arrow::Array>> in,
+    const std::vector<std::shared_ptr<arrow::Array>>& in,
     std::function<arrow::Status(uint64_t, uint64_t)>* on_valid,
     std::function<arrow::Status()>* on_null) {
   return arrow::Status::NotImplemented("ActionBase Submit is abstract.");
@@ -143,7 +143,7 @@ class UniqueAction : public ActionBase {
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -302,7 +302,7 @@ class CountAction : public ActionBase {
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -465,7 +465,7 @@ class CountDistinctAction : public ActionBase {
 #endif
   }
   std::string getName() { return "CountDistinctAction"; }
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -616,7 +616,7 @@ class CountLiteralAction : public ActionBase {
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -740,7 +740,7 @@ class MinAction<DataType, CType, precompile::enable_if_number<DataType>>
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -1022,7 +1022,7 @@ class MinAction<DataType, CType, precompile::enable_if_decimal<DataType>>
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -1179,7 +1179,7 @@ class MinAction<DataType, CType, precompile::enable_if_string_like<DataType>>
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -1350,7 +1350,7 @@ class MaxAction<DataType, CType, precompile::enable_if_number<DataType>>
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -1627,7 +1627,7 @@ class MaxAction<DataType, CType, precompile::enable_if_decimal<DataType>>
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -1783,7 +1783,7 @@ class MaxAction<DataType, CType, precompile::enable_if_string_like<DataType>>
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -1956,7 +1956,7 @@ class SumAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -2117,7 +2117,7 @@ class SumAction<DataType, CType, ResDataType, ResCType,
 
   int RequiredColNum() { return 1; }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -2284,7 +2284,7 @@ class SumActionPartial<DataType, CType, ResDataType, ResCType,
 
   int RequiredColNum() { return 1; }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -2302,7 +2302,7 @@ class SumActionPartial<DataType, CType, ResDataType, ResCType,
     // prepare evaluate lambda
     if (in_null_count_) {
       *on_valid = [this](int dest_group_id) {
-        const bool is_null = in_null_count_ > 0 && in_->IsNull(row_id);
+        const bool is_null = in_->IsNull(row_id);
         if (!is_null) {
           cache_validity_[dest_group_id] = true;
           cache_[dest_group_id] += data_[row_id];
@@ -2450,7 +2450,7 @@ class SumActionPartial<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -2467,7 +2467,7 @@ class SumActionPartial<DataType, CType, ResDataType, ResCType,
     // prepare evaluate lambda
     if (in_null_count_) {
       *on_valid = [this](int dest_group_id) {
-        const bool is_null = in_null_count_ > 0 && in_->IsNull(row_id);
+        const bool is_null = in_->IsNull(row_id);
         if (!is_null) {
           cache_validity_[dest_group_id] = true;
           cache_[dest_group_id] += in_->GetView(row_id);
@@ -2621,7 +2621,7 @@ class AvgAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -2792,7 +2792,7 @@ class AvgAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -2979,7 +2979,7 @@ class SumCountAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -3005,7 +3005,7 @@ class SumCountAction<DataType, CType, ResDataType, ResCType,
       };
     } else {
       *on_valid = [this](int dest_group_id) {
-        const bool is_null = in_null_count_ > 0 && in_->IsNull(row_id);
+        const bool is_null = in_->IsNull(row_id);
         if (!is_null) {
           cache_sum_[dest_group_id] += data_[row_id];
           cache_count_[dest_group_id] += 1;
@@ -3168,7 +3168,7 @@ class SumCountAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -3185,7 +3185,7 @@ class SumCountAction<DataType, CType, ResDataType, ResCType,
     row_id = 0;
     if (in_null_count_) {
       *on_valid = [this](int dest_group_id) {
-        const bool is_null = in_null_count_ > 0 && in_->IsNull(row_id);
+        const bool is_null = in_->IsNull(row_id);
         if (!is_null) {
           cache_sum_[dest_group_id] += in_->GetView(row_id);
           cache_count_[dest_group_id] += 1;
@@ -3353,7 +3353,7 @@ class SumCountMergeAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -3534,7 +3534,7 @@ class SumCountMergeAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -3704,7 +3704,7 @@ class AvgByCountAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -3884,7 +3884,7 @@ class AvgByCountAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -4079,7 +4079,7 @@ class StddevSampPartialAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -4313,7 +4313,7 @@ class StddevSampPartialAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -4540,7 +4540,7 @@ class StddevSampFinalAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -4747,7 +4747,7 @@ class StddevSampFinalAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -4986,7 +4986,7 @@ class FirstPartialAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -5059,8 +5059,9 @@ class FirstPartialAction<DataType, CType, ResDataType, ResCType,
       return arrow::Status::OK();
     }
     auto input_array = std::make_shared<ArrayType>(in[0]);
+    int in_null_count = input_array->null_count();
     for (int id = 0; id < input_array->length(); id++) {
-      if (input_array->IsNull(id)) {
+      if (in_null_count > 0 && input_array->IsNull(id)) {
         if (ignore_nulls_) {
           continue;
         } else {
@@ -5224,7 +5225,7 @@ class FirstPartialAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -5296,8 +5297,9 @@ class FirstPartialAction<DataType, CType, ResDataType, ResCType,
       return arrow::Status::OK();
     }
     auto input_array = std::make_shared<ArrayType>(in[0]);
+    int in_null_count = input_array->null_count();
     for (int id = 0; id < input_array->length(); id++) {
-      if (input_array->IsNull(id)) {
+      if (in_null_count > 0 && input_array->IsNull(id)) {
         if (ignore_nulls_) {
           continue;
         } else {
@@ -5459,7 +5461,7 @@ class FirstFinalAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data
@@ -5719,7 +5721,7 @@ class FirstFinalAction<DataType, CType, ResDataType, ResCType,
 #endif
   }
 
-  arrow::Status Submit(ArrayList in_list, int max_group_id,
+  arrow::Status Submit(const ArrayList& in_list, int max_group_id,
                        std::function<arrow::Status(int)>* on_valid,
                        std::function<arrow::Status()>* on_null) override {
     // resize result data

diff --git a/native-sql-engine/cpp/src/codegen/arrow_compute/ext/actions_impl.h b/native-sql-engine/cpp/src/codegen/arrow_compute/ext/actions_impl.h
@@ -40,10 +40,10 @@ class ActionBase {
  public:
   virtual ~ActionBase() {}
 
-  virtual arrow::Status Submit(ArrayList in, int max_group_id,
+  virtual arrow::Status Submit(const ArrayList& in, int max_group_id,
                                std::function<arrow::Status(int)>* on_valid,
                                std::function<arrow::Status()>* on_null);
-  virtual arrow::Status Submit(std::vector<std::shared_ptr<arrow::Array>> in,
+  virtual arrow::Status Submit(const std::vector<std::shared_ptr<arrow::Array>>& in,
                                std::function<arrow::Status(uint64_t, uint64_t)>* on_valid,
                                std::function<arrow::Status()>* on_null);
   virtual arrow::Status Submit(const std::shared_ptr<arrow::Array>& in,