Remove weird dop logic.

Author: fengttt

go.mod

@@ -14,7 +14,6 @@ require (
 	github.com/aliyun/alibaba-cloud-sdk-go v1.63.34
 	github.com/aliyun/aliyun-oss-go-sdk v3.0.2+incompatible
 	github.com/aliyun/credentials-go v1.3.10
-	github.com/ashvardanian/stringzilla/golang v0.0.0-20251129144314-ad86638a56f7
 	github.com/aws/aws-sdk-go v1.55.5
 	github.com/aws/aws-sdk-go-v2 v1.32.5
 	github.com/aws/aws-sdk-go-v2/config v1.28.5

go.sum

@@ -68,8 +68,6 @@ github.com/andybalholm/brotli v1.1.0/go.mod h1:sms7XGricyQI9K10gOSf56VKKWS4oLer5
 github.com/armon/consul-api v0.0.0-20180202201655-eb2c6b5be1b6/go.mod h1:grANhF5doyWs3UAsr3K4I6qtAmlQcZDesFNEHPZAzj8=
 github.com/armon/go-metrics v0.0.0-20180917152333-f0300d1749da h1:8GUt8eRujhVEGZFFEjBj46YV4rDjvGrNxb0KMWYkL2I=
 github.com/armon/go-metrics v0.0.0-20180917152333-f0300d1749da/go.mod h1:Q73ZrmVTwzkszR9V5SSuryQ31EELlFMUz1kKyl939pY=
-github.com/ashvardanian/stringzilla/golang v0.0.0-20251129144314-ad86638a56f7 h1:JIEbjZ/eWa9i+l3W2/KNh4LS33rKTZLH9Q797y+urhg=
-github.com/ashvardanian/stringzilla/golang v0.0.0-20251129144314-ad86638a56f7/go.mod h1:iQfeN5MGwtAidmePxLg0w7DsKRNA3L83OHfkCYMBPEQ=
 github.com/aws/aws-sdk-go v1.55.5 h1:KKUZBfBoyqy5d3swXyiC7Q76ic40rYcbqH7qjh59kzU=
 github.com/aws/aws-sdk-go v1.55.5/go.mod h1:eRwEWoyTWFMVYVQzKMNHWP5/RV4xIUGMQfXQHfHkpNU=
 github.com/aws/aws-sdk-go-v2 v1.32.5 h1:U8vdWJuY7ruAkzaOdD7guwJjD06YSKmnKCJs7s3IkIo=

pkg/container/hashtable/common.go

@@ -37,18 +37,6 @@ func maxElemCnt(cellCnt, cellSize uint64) uint64 {
 	return cellCnt * 4 / 5
 }
 
-type Aggregator interface {
-	StateSize() uint8
-	ResultSize() uint8
-	NeedsInit() bool
-
-	Init(state unsafe.Pointer)
-	AddBatch(states []unsafe.Pointer, values unsafe.Pointer)
-	MergeBatch(lstates, rstates []unsafe.Pointer)
-
-	Finalize(states, results []unsafe.Pointer)
-}
-
 func toUnsafePointer[T any](p *T) unsafe.Pointer {
 	return unsafe.Pointer(p)
 }

pkg/sql/plan/shuffle.go

@@ -677,43 +677,6 @@ func determineShuffleForGroupBy(node *plan.Node, builder *QueryBuilder) {
 
 }
 
-func getShuffleDop(ncpu int, lencn int, hashmapSize float64) (dop int) {
-	if ncpu <= 4 {
-		ncpu = 4
-	}
-	maxret := ncpu * 4
-	// these magic number comes from hashmap resize factor. see hashtable/common.go, in maxElemCnt function
-	ret1 := int(hashmapSize/float64(lencn)/12800000) + 1
-	if ret1 >= maxret {
-		return maxret
-	}
-
-	ret2 := int(hashmapSize/float64(lencn)/6000000) + 1
-	if ret2 >= maxret {
-		return ret1
-	}
-	ret3 := int(hashmapSize/float64(lencn)/320000) + 1
-	if ret3 <= ncpu/2 {
-		return ncpu
-	}
-	if ret3 >= maxret-1 {
-		return maxret
-	}
-	if ret3 <= ncpu {
-		if ncpu*2 > maxret {
-			return maxret
-		} else {
-			return ncpu * 2
-		}
-	}
-	ret4 := (ret3/ncpu + 1) * ncpu
-	if ret4 > maxret {
-		return maxret
-	} else {
-		return ret4
-	}
-}
-
 // default shuffle type for scan is hash
 // for table with primary key, and ndv of first column in primary key is high enough, use range shuffle
 // only support integer type

pkg/sql/plan/shuffle_test.go

@@ -203,69 +203,6 @@ func TestRangeShuffleSlice(t *testing.T) {
 	require.Equal(t, GetRangeShuffleIndexUnsignedSlice([]uint64{30, 50, 60, 90, 120}, 61), uint64(3))
 }
 
-func TestGetShuffleDop(t *testing.T) {
-	n := getShuffleDop(1, 1, 100)
-	require.Equal(t, 4, n)
-	n = getShuffleDop(4, 2, 100000000000)
-	require.Equal(t, 16, n)
-	n = getShuffleDop(64, 1, 10000000000000)
-	require.Equal(t, 256, n)
-
-	n = getShuffleDop(16, 3, 1500000000)
-	require.Equal(t, 40, n)
-	n = getShuffleDop(16, 3, 150000000)
-	require.Equal(t, 64, n)
-	n = getShuffleDop(16, 3, 15000000)
-	require.Equal(t, 32, n)
-	n = getShuffleDop(16, 3, 1500000)
-	require.Equal(t, 16, n)
-
-	n = getShuffleDop(16, 4, 1500000000)
-	require.Equal(t, 64, n)
-	n = getShuffleDop(16, 4, 150000000)
-	require.Equal(t, 64, n)
-	n = getShuffleDop(16, 4, 15000000)
-	require.Equal(t, 32, n)
-	n = getShuffleDop(16, 4, 1500000)
-	require.Equal(t, 16, n)
-
-	n = getShuffleDop(16, 3, 300000000)
-	require.Equal(t, 64, n)
-	n = getShuffleDop(16, 3, 30000000)
-	require.Equal(t, 48, n)
-	n = getShuffleDop(16, 3, 3000000)
-	require.Equal(t, 16, n)
-	n = getShuffleDop(16, 3, 300000)
-	require.Equal(t, 16, n)
-
-	n = getShuffleDop(16, 4, 300000000)
-	require.Equal(t, 64, n)
-	n = getShuffleDop(16, 4, 30000000)
-	require.Equal(t, 32, n)
-	n = getShuffleDop(16, 4, 3000000)
-	require.Equal(t, 16, n)
-	n = getShuffleDop(16, 4, 300000)
-	require.Equal(t, 16, n)
-
-	n = getShuffleDop(64, 1, 1500000000)
-	require.Equal(t, 256, n)
-	n = getShuffleDop(64, 1, 150000000)
-	require.Equal(t, 256, n)
-	n = getShuffleDop(64, 1, 15000000)
-	require.Equal(t, 128, n)
-	n = getShuffleDop(64, 1, 1500000)
-	require.Equal(t, 64, n)
-
-	n = getShuffleDop(16, 1, 1500000000)
-	require.Equal(t, 64, n)
-	n = getShuffleDop(16, 1, 150000000)
-	require.Equal(t, 64, n)
-	n = getShuffleDop(16, 1, 15000000)
-	require.Equal(t, 48, n)
-	n = getShuffleDop(16, 1, 1500000)
-	require.Equal(t, 16, n)
-}
-
 func TestShouldSkipObjByShuffle(t *testing.T) {
 	row := types.RandomRowid()
 	stats := objectio.NewObjectStatsWithObjectID(row.BorrowObjectID(), false, false, true)

pkg/sql/plan/stats.go

@@ -1872,7 +1872,9 @@ func CalcNodeDOP(p *plan.Plan, rootID int32, ncpu int32, lencn int) {
 		if node.NodeType == plan.Node_JOIN && node.JoinType == plan.Node_DEDUP {
 			setNodeDOP(p, rootID, ncpu)
 		} else {
-			dop := int32(getShuffleDop(int(ncpu), lencn, node.Stats.HashmapStats.HashmapSize))
+			// there was a weird calculate shuffle dop logic here, which does not really make any sense.
+			// just use ncpu.
+			dop := ncpu
 			childDop := qry.Nodes[node.Children[0]].Stats.Dop
 			if dop < childDop {
 				dop = childDop