prefetch threshold(2MB); mini batch; prefetch exists key; prefetch in…

…sertResultInto Signed-off-by: guo-shaoge <[email protected]>
pingcap · Dec 24, 2024 · 9623368 · 9623368
1 parent dc0973a
commit 9623368
Show file tree

Hide file tree

Showing 4 changed files with 244 additions and 70 deletions.
diff --git a/dbms/src/AggregateFunctions/IAggregateFunction.h b/dbms/src/AggregateFunctions/IAggregateFunction.h
@@ -236,40 +236,25 @@ class IAggregateFunctionHelper : public IAggregateFunction
         ssize_t if_argument_pos = -1) const override
     {
         const auto end = start_offset + batch_size;
-        static constexpr size_t prefetch_step = 16;
         if (if_argument_pos >= 0)
         {
             const auto & flags = assert_cast<const ColumnUInt8 &>(*columns[if_argument_pos]).getData();
             for (size_t i = start_offset; i < end; ++i)
             {
                 const auto place_idx = i - start_offset;
-                const auto prefetch_idx = place_idx + prefetch_step;
 
                 if (flags[i] && places[place_idx])
-                {
-                    // Only prefetch first agg function data for simplicity.
-                    // This may not work when there are too many agg functions.
-                    if likely (prefetch_idx < end)
-                        __builtin_prefetch(places[prefetch_idx]);
-
                     static_cast<const Derived *>(this)->add(places[place_idx] + place_offset, columns, i, arena);
-                }
             }
         }
         else
         {
             for (size_t i = start_offset; i < end; ++i)
             {
                 const auto place_idx = i - start_offset;
-                const auto prefetch_idx = place_idx + prefetch_step;
 
                 if (places[place_idx])
-                {
-                    if likely (prefetch_idx < end)
-                        __builtin_prefetch(places[prefetch_idx]);
-
                     static_cast<const Derived *>(this)->add(places[place_idx] + place_offset, columns, i, arena);
-                }
             }
         }
     }

diff --git a/dbms/src/Common/ColumnsHashingImpl.h b/dbms/src/Common/ColumnsHashingImpl.h
@@ -150,6 +150,18 @@ class HashMethodBase
         return emplaceImpl(key_holder, data);
     }
 
+    template <typename Data>
+    ALWAYS_INLINE inline EmplaceResult emplaceKey(
+        Data & data,
+        size_t row,
+        Arena & pool,
+        std::vector<String> & sort_key_containers,
+        size_t hashval)
+    {
+        auto key_holder = static_cast<Derived &>(*this).getKeyHolder(row, &pool, sort_key_containers);
+        return emplaceImpl(key_holder, data, hashval);
+    }
+
     template <typename Data>
     ALWAYS_INLINE inline FindResult findKey(
         Data & data,
@@ -161,6 +173,18 @@ class HashMethodBase
         return findKeyImpl(keyHolderGetKey(key_holder), data);
     }
 
+    template <typename Data>
+    ALWAYS_INLINE inline FindResult findKey(
+        Data & data,
+        size_t row,
+        Arena & pool,
+        std::vector<String> & sort_key_containers,
+        size_t hashval)
+    {
+        auto key_holder = static_cast<Derived &>(*this).getKeyHolder(row, &pool, sort_key_containers);
+        return findKeyImpl(keyHolderGetKey(key_holder), data, hashval);
+    }
+
     // Emplace key using hashval, you can enable prefetch or not.
     template <bool enable_prefetch = false, typename Data>
     ALWAYS_INLINE inline EmplaceResult emplaceKey(