address cr.

Author: Rachelint

datafusion/physical-plan/src/aggregates/group_values/single_group_by/primitive.rs

@@ -81,10 +81,13 @@ hash_float!(f16, f32, f64);
 pub struct GroupValuesPrimitive<T: ArrowPrimitiveType> {
     /// The data type of the output array
     data_type: DataType,
-    /// Stores the group index based on the hash of its value
+    /// Stores the `(group_index, hash)` based on the hash of its value
+    ///
+    /// We also store `hash` is for reducing cost of rehashing. Such cost
+    /// is obvious in high cardinality group by situation.
+    /// More details can see:
+    /// <https://github.com/apache/datafusion/issues/15961>
     ///
-    /// We don't store the hashes as hashing fixed width primitives
-    /// is fast enough for this not to benefit performance
     map: HashTable<(usize, u64)>,
     /// The group index of the null value if any
     null_group: Option<usize>,
@@ -148,7 +151,7 @@ where
     }
 
     fn size(&self) -> usize {
-        self.map.capacity() * size_of::<usize>() + self.values.allocated_size()
+        self.map.capacity() * size_of::<(usize, u64)>() + self.values.allocated_size()
     }
 
     fn is_empty(&self) -> bool {