Additional 10% perf on btree inserts

dasatomic · Dec 7, 2021 · eac10e2 · eac10e2
1 parent 10745c2
commit eac10e2
Show file tree

Hide file tree

Showing 43 changed files with 350 additions and 37 deletions.
diff --git a/DataStructures/BTreeCollection.cs b/DataStructures/BTreeCollection.cs
@@ -186,7 +186,7 @@ public async Task Add(RowHolder item, ITransaction tran)
                     {
                         this.rowUniqueCheckPageLevel(currPage, itemToInsert, tran);
 
-                        int pos = currPage.InsertOrdered(itemToInsert, tran, this.btreeColumnTypes, this.indexComparer);
+                        int pos = currPage.InsertOrdered(itemToInsert, tran, this.btreeColumnTypes, this.indexPosition);
                         Debug.Assert(pos >= 0);
 
                         if (debugPrintPage != null)
@@ -209,12 +209,12 @@ public async Task Add(RowHolder item, ITransaction tran)
                         if (compareResult < 0)
                         {
                             // left.
-                            int pos = currPage.InsertOrdered(itemToInsert, tran, this.btreeColumnTypes, this.indexComparer);
+                            int pos = currPage.InsertOrdered(itemToInsert, tran, this.btreeColumnTypes, this.indexPosition);
                         }
                         else
                         {
                             // right.
-                            int pos = newPageForSplit.InsertOrdered(itemToInsert, tran, this.btreeColumnTypes, this.indexComparer);
+                            int pos = newPageForSplit.InsertOrdered(itemToInsert, tran, this.btreeColumnTypes, this.indexPosition);
                         }
 
                         insertFinished = true;
@@ -421,7 +421,7 @@ private async Task<RowHolder> SplitBtreePage(MixedPage currPage, MixedPage newPa
 
             if (prevPage != null)
             {
-                int pos = prevPage.InsertOrdered(rowHolderForSplit, tran, this.btreeColumnTypes, this.indexComparer);
+                int pos = prevPage.InsertOrdered(rowHolderForSplit, tran, this.btreeColumnTypes, this.indexPosition);
 
                 // We know that there will be enough space since we proactivly clean parent nodes.
                 Debug.Assert(pos >= 0);
@@ -435,7 +435,7 @@ private async Task<RowHolder> SplitBtreePage(MixedPage currPage, MixedPage newPa
                 MixedPage newRoot = await pageAllocator.AllocateMixedPage(this.btreeColumnTypes, PageManagerConstants.NullPageId, PageManagerConstants.NullPageId, tran);
                 this.SetLeaf(newRoot, false);
                 using Releaser newRootLock = await tran.AcquireLock(newPageForSplit.PageId(), LockManager.LockTypeEnum.Exclusive).ConfigureAwait(false);
-                int pos = newRoot.InsertOrdered(rowHolderForSplit, tran, this.btreeColumnTypes, this.indexComparer);
+                int pos = newRoot.InsertOrdered(rowHolderForSplit, tran, this.btreeColumnTypes, this.indexPosition);
                 Debug.Assert(pos == 0);
 
                 newRoot.SetPrevPageId(currPage.PageId());

diff --git a/PageManager/PageTypes/IPageSerializer.cs b/PageManager/PageTypes/IPageSerializer.cs
@@ -66,7 +66,7 @@ public void ResetDirty()
 
         public abstract void Update(ST item, ushort position, ITransaction transaction);
         public abstract int Insert(ST item, ITransaction transaction);
-        public abstract int InsertOrdered(ST item, ITransaction transaction, ColumnInfo[] columnTypes, Func<RowHolder, RowHolder, int> comparer);
+        public abstract int InsertOrdered(ST item, ITransaction transaction, ColumnInfo[] columnTypes, int comparisonField);
         public abstract void At(ushort position, ITransaction tran, ref ST item);
         public abstract void SplitPage(MixedPage newPage, ref RowHolder splitValue, int elemNumForSplit, ITransaction transaction);
 

diff --git a/PageManager/PageTypes/MixedPageType.cs b/PageManager/PageTypes/MixedPageType.cs
@@ -133,13 +133,13 @@ public override int Insert(RowHolder item, ITransaction transaction)
             }
         }
 
-        public override int InsertOrdered(RowHolder item, ITransaction transaction, ColumnInfo[] columnTypes, Func<RowHolder, RowHolder, int> comparer)
+        public override int InsertOrdered(RowHolder item, ITransaction transaction, ColumnInfo[] columnTypes, int comparisonField)
         {
             transaction.VerifyLock(this.pageId, LockManager.LockTypeEnum.Exclusive);
 
             lock (this.lockObject)
             {
-                int position = this.items.InsertRowOrdered(item, columnTypes, comparer);
+                int position = this.items.InsertRowOrdered(item, columnTypes, comparisonField);
 
                 if (position == -1)
                 {

diff --git a/PageManager/PageTypes/StringOnlyPage.cs b/PageManager/PageTypes/StringOnlyPage.cs
@@ -322,7 +322,7 @@ public override void At(ushort position, ITransaction tran, ref char[] item)
             throw new NotImplementedException();
         }
 
-        public override int InsertOrdered(char[] item, ITransaction transaction, ColumnInfo[] columnTypes, Func<RowHolder, RowHolder, int> comparer)
+        public override int InsertOrdered(char[] item, ITransaction transaction, ColumnInfo[] columnTypes, int comparisonField) 
         {
             throw new NotImplementedException();
         }

diff --git a/PageManager/RowsetHolder.cs b/PageManager/RowsetHolder.cs
@@ -36,6 +36,8 @@ public unsafe struct RowsetHolder
 
         private ushort rowCount;
 
+        private static readonly CompareWithRowHolderCreator compareWithRowHolderCreator = new CompareWithRowHolderCreator();
+
         public RowsetHolder(ColumnInfo[] columnTypes, Memory<byte> storage, bool init)
         {
             System.Diagnostics.Debug.Assert(BitConverter.IsLittleEndian, "Rowset holder fixed assumes that we are running on little endian");
@@ -106,6 +108,13 @@ public void UpdateRowCount()
             }
         }
 
+        public int CompareFieldWithRowHolder(int row, int col, RowHolder rh, ColumnInfo ci)
+        {
+            return ColumnTypeHandlerRouter<Func<int, int, RowHolder, ColumnInfo, RowsetHolder, int>>.Route(
+                compareWithRowHolderCreator,
+                ci.ColumnType)(row, col, rh, ci, this);
+        }
+
         public void GetRow(int row, ref RowHolder rowHolder)
         {
             System.Diagnostics.Debug.Assert(IsPresent(row));
@@ -143,9 +152,10 @@ public int InsertRow(RowHolder rowHolder)
             }
         }
 
-        public int InsertRowOrdered(RowHolder rowHolderToInsert, ColumnInfo[] columnTypes, Func<RowHolder, RowHolder, int> comparer)
+        public int InsertRowOrdered(RowHolder rowHolderToInsert, ColumnInfo[] columnTypes, int comparisonField)
         {
             // find the first element that is bigger than one to insert.
+            // No need to keep free items. Try to compact every time.
             // TODO: this can be logn.
             int positionToInsert = -1;
             bool insertAtEnd = false;
@@ -154,11 +164,8 @@ public int InsertRowOrdered(RowHolder rowHolderToInsert, ColumnInfo[] columnType
             {
                 if (BitArray.IsSet(i, this.storage.Span))
                 {
-                    RowHolder rowHolder = new RowHolder(columnTypes);
-                    GetRow(i, ref rowHolder);
-                    if (comparer(rowHolderToInsert, rowHolder) != 1)
+                    if (this.CompareFieldWithRowHolder(i, comparisonField, rowHolderToInsert, columnTypes[comparisonField]) != -1)
                     {
-                        // I am bigger than you, I should be at your place.
                         positionToInsert = i;
                         break;
                     }
@@ -445,4 +452,34 @@ private int GetTuplePosition(int row, int col)
             return tuplePosition + offsetInTouple;
         }
     }
+
+    public class CompareWithRowHolderCreator : ColumnTypeHandlerBasicSingle<Func<int, int, RowHolder, ColumnInfo, RowsetHolder, int>>
+    {
+        public Func<int, int, RowHolder, ColumnInfo, RowsetHolder, int> HandleDouble()
+        {
+            return (row, col, rh, ci, rs) =>
+            {
+                double fieldRowsetHolder = rs.GetRowGeneric<double>(row, col);
+                double fieldFromRowHolder = rh.GetField<double>(col);
+
+                return fieldRowsetHolder.CompareTo(fieldFromRowHolder);
+            };
+        }
+
+        public Func<int, int, RowHolder, ColumnInfo, RowsetHolder, int> HandleInt()
+        {
+            return (row, col, rh, ci, rs) =>
+            {
+                int fieldRowsetHolder = rs.GetRowGeneric<int>(row, col);
+                int fieldFromRowHolder = rh.GetField<int>(col);
+
+                return fieldRowsetHolder.CompareTo(fieldFromRowHolder);
+            };
+        }
+
+        public Func<int, int, RowHolder, ColumnInfo, RowsetHolder, int> HandleString()
+        {
+            throw new NotImplementedException();
+        }
+    }
 }
diff --git a/...chmarkDotNet.Artifacts/BTreeInsertPerf.cs → UnitBenchmark/BTreeInsertPerf.cs b/...chmarkDotNet.Artifacts/BTreeInsertPerf.cs → UnitBenchmark/BTreeInsertPerf.cs
diff --git a/...toBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=10000-cummean.png b/...toBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=10000-cummean.png
diff --git a/...toBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=10000-density.png b/...toBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=10000-density.png
diff --git a/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=10000-timeline.png b/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=10000-timeline.png
diff --git a/...SingleIntColumnRandomData-DefaultJob RowsInTableNumber=10000-timelineSmooth.png b/...SingleIntColumnRandomData-DefaultJob RowsInTableNumber=10000-timelineSmooth.png
diff --git a/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=100000-cummean.png b/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=100000-cummean.png
diff --git a/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=100000-density.png b/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=100000-density.png
diff --git a/...BTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=100000-timeline.png b/...BTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=100000-timeline.png
diff --git a/...ingleIntColumnRandomData-DefaultJob RowsInTableNumber=100000-timelineSmooth.png b/...ingleIntColumnRandomData-DefaultJob RowsInTableNumber=100000-timelineSmooth.png
diff --git a/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=200000-cummean.png b/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=200000-cummean.png
diff --git a/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=200000-density.png b/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=200000-density.png
diff --git a/...BTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=200000-timeline.png b/...BTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=200000-timeline.png
diff --git a/...ingleIntColumnRandomData-DefaultJob RowsInTableNumber=200000-timelineSmooth.png b/...ingleIntColumnRandomData-DefaultJob RowsInTableNumber=200000-timelineSmooth.png
diff --git a/...toBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=50000-cummean.png b/...toBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=50000-cummean.png
diff --git a/...toBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=50000-density.png b/...toBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=50000-density.png
diff --git a/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=50000-timeline.png b/...oBTreeSingleIntColumnRandomData-DefaultJob RowsInTableNumber=50000-timeline.png
diff --git a/...SingleIntColumnRandomData-DefaultJob RowsInTableNumber=50000-timelineSmooth.png b/...SingleIntColumnRandomData-DefaultJob RowsInTableNumber=50000-timelineSmooth.png
diff --git a/...rf-InsertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=10000-density.png b/...rf-InsertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=10000-density.png
diff --git a/...sertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=10000-facetDensity.png b/...sertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=10000-facetDensity.png
diff --git a/...f-InsertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=100000-density.png b/...f-InsertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=100000-density.png
diff --git a/...ertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=100000-facetDensity.png b/...ertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=100000-facetDensity.png
diff --git a/...f-InsertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=200000-density.png b/...f-InsertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=200000-density.png
diff --git a/...ertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=200000-facetDensity.png b/...ertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=200000-facetDensity.png
diff --git a/...rf-InsertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=50000-density.png b/...rf-InsertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=50000-density.png
diff --git a/...sertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=50000-facetDensity.png b/...sertIntoBTreeSingleIntColumnRandomData-RowsInTableNumber=50000-facetDensity.png
diff --git a/...tBenchmark.BTreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-cummean.png b/...tBenchmark.BTreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-cummean.png
diff --git a/...tBenchmark.BTreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-density.png b/...tBenchmark.BTreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-density.png
diff --git a/...hmark.BTreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-facetDensity.png b/...hmark.BTreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-facetDensity.png
diff --git a/...mark.BTreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-facetTimeline.png b/...mark.BTreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-facetTimeline.png
diff --git a/...TreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-facetTimelineSmooth.png b/...TreeInsertPerf-InsertIntoBTreeSingleIntColumnRandomData-facetTimelineSmooth.png
diff --git a/...ark/BenchmarkDotNet.Artifacts/results/UnitBenchmark.BTreeInsertPerf-barplot.png b/...ark/BenchmarkDotNet.Artifacts/results/UnitBenchmark.BTreeInsertPerf-barplot.png
diff --git a/...ark/BenchmarkDotNet.Artifacts/results/UnitBenchmark.BTreeInsertPerf-boxplot.png b/...ark/BenchmarkDotNet.Artifacts/results/UnitBenchmark.BTreeInsertPerf-boxplot.png