Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

planner: consider agg func type in cost model #12038

Merged
merged 3 commits into from
Sep 9, 2019
Merged

planner: consider agg func type in cost model #12038

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
261f6a5
consider agg func type in cost model
lzmhhh123 Sep 5, 2019
8c46d76
address comment
lzmhhh123 Sep 6, 2019
bed80af
Merge branch 'master' into fix_cost_model
sre-bot Sep 9, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 23 additions & 30 deletions cmd/explaintest/r/explain_easy.result
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -98,14 +98,13 @@ StreamAgg_12 1.00 root funcs:sum(col_0)
explain select c1 from t1 where c1 in (select c2 from t2);
id count task operator info
Projection_9 9990.00 root test.t1.c1
└─HashLeftJoin_17 9990.00 root inner join, inner:HashAgg_24, equal:[eq(test.t1.c1, test.t2.c2)]
└─HashLeftJoin_17 9990.00 root inner join, inner:HashAgg_21, equal:[eq(test.t1.c1, test.t2.c2)]
├─TableReader_30 10000.00 root data:TableScan_29
│ └─TableScan_29 10000.00 cop table:t1, range:[-inf,+inf], keep order:false, stats:pseudo
└─HashAgg_24 7992.00 root group by:col_1, funcs:firstrow(col_1)
└─TableReader_25 7992.00 root data:HashAgg_19
└─HashAgg_19 7992.00 cop group by:test.t2.c2,
└─Selection_23 9990.00 cop not(isnull(test.t2.c2))
└─TableScan_22 10000.00 cop table:t2, range:[-inf,+inf], keep order:false, stats:pseudo
└─HashAgg_21 7992.00 root group by:test.t2.c2, funcs:firstrow(test.t2.c2)
└─TableReader_28 9990.00 root data:Selection_27
└─Selection_27 9990.00 cop not(isnull(test.t2.c2))
└─TableScan_26 10000.00 cop table:t2, range:[-inf,+inf], keep order:false, stats:pseudo
explain select (select count(1) k from t1 s where s.c1 = t1.c1 having k != 0) from t1;
id count task operator info
Projection_12 10000.00 root ifnull(5_col_0, 0)
Expand Down Expand Up @@ -165,32 +164,27 @@ id count task operator info
Union_17 26000.00 root
├─HashAgg_21 16000.00 root group by:c1, funcs:firstrow(join_agg_0)
│ └─Union_22 16000.00 root
│ ├─StreamAgg_34 8000.00 root group by:col_2, funcs:firstrow(col_2), firstrow(col_2)
│ │ └─IndexReader_35 8000.00 root index:StreamAgg_26
│ │ └─StreamAgg_26 8000.00 cop group by:test.t2.c1,
│ │ └─IndexScan_33 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
│ └─StreamAgg_49 8000.00 root group by:col_2, funcs:firstrow(col_2), firstrow(col_2)
│ └─IndexReader_50 8000.00 root index:StreamAgg_41
│ └─StreamAgg_41 8000.00 cop group by:test.t2.c1,
│ └─IndexScan_48 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
│ ├─StreamAgg_27 8000.00 root group by:test.t2.c1, funcs:firstrow(test.t2.c1), firstrow(test.t2.c1)
│ │ └─IndexReader_37 10000.00 root index:IndexScan_36
│ │ └─IndexScan_36 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
│ └─StreamAgg_42 8000.00 root group by:test.t2.c1, funcs:firstrow(test.t2.c1), firstrow(test.t2.c1)
│ └─IndexReader_52 10000.00 root index:IndexScan_51
│ └─IndexScan_51 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
└─TableReader_55 10000.00 root data:TableScan_54
└─TableScan_54 10000.00 cop table:t2, range:[-inf,+inf], keep order:false, stats:pseudo
explain select c1 from t2 union all select c1 from t2 union select c1 from t2;
id count task operator info
HashAgg_18 24000.00 root group by:c1, funcs:firstrow(join_agg_0)
└─Union_19 24000.00 root
├─StreamAgg_31 8000.00 root group by:col_2, funcs:firstrow(col_2), firstrow(col_2)
│ └─IndexReader_32 8000.00 root index:StreamAgg_23
│ └─StreamAgg_23 8000.00 cop group by:test.t2.c1,
│ └─IndexScan_30 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
├─StreamAgg_46 8000.00 root group by:col_2, funcs:firstrow(col_2), firstrow(col_2)
│ └─IndexReader_47 8000.00 root index:StreamAgg_38
│ └─StreamAgg_38 8000.00 cop group by:test.t2.c1,
│ └─IndexScan_45 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
└─StreamAgg_61 8000.00 root group by:col_2, funcs:firstrow(col_2), firstrow(col_2)
└─IndexReader_62 8000.00 root index:StreamAgg_53
└─StreamAgg_53 8000.00 cop group by:test.t2.c1,
└─IndexScan_60 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
├─StreamAgg_24 8000.00 root group by:test.t2.c1, funcs:firstrow(test.t2.c1), firstrow(test.t2.c1)
│ └─IndexReader_34 10000.00 root index:IndexScan_33
│ └─IndexScan_33 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
├─StreamAgg_39 8000.00 root group by:test.t2.c1, funcs:firstrow(test.t2.c1), firstrow(test.t2.c1)
│ └─IndexReader_49 10000.00 root index:IndexScan_48
│ └─IndexScan_48 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
└─StreamAgg_54 8000.00 root group by:test.t2.c1, funcs:firstrow(test.t2.c1), firstrow(test.t2.c1)
└─IndexReader_64 10000.00 root index:IndexScan_63
└─IndexScan_63 10000.00 cop table:t2, index:c1, range:[NULL,+inf], keep order:true, stats:pseudo
explain select count(1) from (select count(1) from (select * from t1 where c3 = 100) k) k2;
id count task operator info
StreamAgg_13 1.00 root funcs:count(1)
Expand All @@ -216,10 +210,9 @@ StreamAgg_11 1.00 root funcs:count(1)
explain select count(1) from (select count(c2) from t1 group by c3) k;
id count task operator info
StreamAgg_11 1.00 root funcs:count(1)
└─HashAgg_23 8000.00 root group by:col_1, funcs:firstrow(col_0)
└─TableReader_24 8000.00 root data:HashAgg_20
└─HashAgg_20 8000.00 cop group by:test.t1.c3, funcs:firstrow(1)
└─TableScan_15 10000.00 cop table:t1, range:[-inf,+inf], keep order:false, stats:pseudo
└─HashAgg_22 8000.00 root group by:test.t1.c3, funcs:firstrow(1)
└─TableReader_19 10000.00 root data:TableScan_18
└─TableScan_18 10000.00 cop table:t1, range:[-inf,+inf], keep order:false, stats:pseudo
set @@session.tidb_opt_insubq_to_join_and_agg=0;
explain select sum(t1.c1 in (select c1 from t2)) from t1;
id count task operator info
Expand Down
11 changes: 5 additions & 6 deletions cmd/explaintest/r/explain_easy_stats.result
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -64,14 +64,13 @@ IndexLookUp_9 0.00 root
explain select count(b.c2) from t1 a, t2 b where a.c1 = b.c2 group by a.c1;
id count task operator info
Projection_11 1985.00 root cast(join_agg_0)
└─HashLeftJoin_19 1985.00 root inner join, inner:HashAgg_26, equal:[eq(test.a.c1, test.b.c2)]
└─HashLeftJoin_19 1985.00 root inner join, inner:HashAgg_23, equal:[eq(test.a.c1, test.b.c2)]
├─TableReader_32 1999.00 root data:TableScan_31
│ └─TableScan_31 1999.00 cop table:a, range:[-inf,+inf], keep order:false
└─HashAgg_26 1985.00 root group by:col_2, funcs:count(col_0), firstrow(col_2)
└─TableReader_27 1985.00 root data:HashAgg_21
└─HashAgg_21 1985.00 cop group by:test.b.c2, funcs:count(test.b.c2)
└─Selection_25 1985.00 cop not(isnull(test.b.c2))
└─TableScan_24 1985.00 cop table:b, range:[-inf,+inf], keep order:false
└─HashAgg_23 1985.00 root group by:test.b.c2, funcs:count(test.b.c2), firstrow(test.b.c2)
└─TableReader_30 1985.00 root data:Selection_29
└─Selection_29 1985.00 cop not(isnull(test.b.c2))
└─TableScan_28 1985.00 cop table:b, range:[-inf,+inf], keep order:false
explain select * from t2 order by t2.c2 limit 0, 1;
id count task operator info
TopN_7 1.00 root test.t2.c2:asc, offset:0, count:1
Expand Down
7 changes: 3 additions & 4 deletions cmd/explaintest/r/index_join.result
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,7 +50,6 @@ Projection_8 10000.00 root test.t1.a, test.t1.b
├─IndexLookUp_11 10.00 root
│ ├─IndexScan_9 10.00 cop table:t1, index:a, range: decided by [eq(test.t1.a, test.t2.a)], keep order:false, stats:pseudo
│ └─TableScan_10 10.00 cop table:t1, keep order:false, stats:pseudo
└─StreamAgg_29 8000.00 root group by:col_1, funcs:firstrow(col_1)
└─IndexReader_30 8000.00 root index:StreamAgg_21
└─StreamAgg_21 8000.00 cop group by:test.t2.a,
└─IndexScan_28 10000.00 cop table:t2, index:a, range:[NULL,+inf], keep order:true, stats:pseudo
└─StreamAgg_22 8000.00 root group by:test.t2.a, funcs:firstrow(test.t2.a)
└─IndexReader_32 10000.00 root index:IndexScan_31
└─IndexScan_31 10000.00 cop table:t2, index:a, range:[NULL,+inf], keep order:true, stats:pseudo
2 changes: 1 addition & 1 deletion planner/core/cbo_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -411,7 +411,7 @@ func (s *testAnalyzeSuite) TestEmptyTable(c *C) {
},
{
sql: "select * from t where c1 in (select c1 from t1)",
best: "LeftHashJoin{TableReader(Table(t)->Sel([not(isnull(test.t.c1))]))->TableReader(Table(t1)->Sel([not(isnull(test.t1.c1))])->HashAgg)->HashAgg}(test.t.c1,test.t1.c1)->Projection",
best: "LeftHashJoin{TableReader(Table(t)->Sel([not(isnull(test.t.c1))]))->TableReader(Table(t1)->Sel([not(isnull(test.t1.c1))]))->HashAgg}(test.t.c1,test.t1.c1)->Projection",
},
{
sql: "select * from t, t1 where t.c1 = t1.c1",
Expand Down
19 changes: 19 additions & 0 deletions planner/core/find_best_task.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@ package core
import (
"math"

"github.com/pingcap/parser/ast"
"github.com/pingcap/parser/model"
"github.com/pingcap/parser/mysql"
"github.com/pingcap/tidb/expression"
Expand Down Expand Up @@ -43,6 +44,24 @@ const (
distinctFactor = 0.8
)

var aggFuncFactor = map[string]float64{
ast.AggFuncCount: 1.0,
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How do you dicide their values?
And what is defalut for?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

  1. The values are how they keep the result of their arguments. For example:
avg: avg.sum += expr, avg.cnt += 1
var_pop: var_pop.sum_sqr += expr*expr, var_pop.cnt += 1, var_pop.sum += expr
  1. The default is just for safety.

ast.AggFuncSum: 1.0,
ast.AggFuncAvg: 2.0,
ast.AggFuncFirstRow: 0.1,
ast.AggFuncMax: 1.0,
ast.AggFuncMin: 1.0,
ast.AggFuncGroupConcat: 1.0,
ast.AggFuncBitOr: 0.9,
ast.AggFuncBitXor: 0.9,
ast.AggFuncBitAnd: 0.9,
ast.AggFuncVarPop: 3.0,
ast.AggFuncVarSamp: 3.0,
ast.AggFuncStddevPop: 3.0,
ast.AggFuncStddevSamp: 3.0,
"default": 1.5,
}

// wholeTaskTypes records all possible kinds of task that a plan can return. For Agg, TopN and Limit, we will try to get
// these tasks one by one.
var wholeTaskTypes = [...]property.TaskType{property.CopSingleReadTaskType, property.CopDoubleReadTaskType, property.RootTaskType}
Expand Down
6 changes: 3 additions & 3 deletions planner/core/physical_plan_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -851,7 +851,7 @@ func (s *testPlanSuite) TestDAGPlanBuilderAgg(c *C) {
// Test distinct.
{
sql: "select distinct b from t",
best: "TableReader(Table(t)->HashAgg)->HashAgg",
best: "TableReader(Table(t))->HashAgg",
},
{
sql: "select count(*) from (select * from t order by b) t group by b",
Expand All @@ -864,7 +864,7 @@ func (s *testPlanSuite) TestDAGPlanBuilderAgg(c *C) {
// Test agg + table.
{
sql: "select sum(a), avg(b + c) from t group by d",
best: "TableReader(Table(t))->Projection->HashAgg",
best: "TableReader(Table(t)->HashAgg)->HashAgg",
},
{
sql: "select sum(distinct a), avg(b + c) from t group by d",
Expand Down Expand Up @@ -1635,7 +1635,7 @@ func (s *testPlanSuite) TestAggregationHints(c *C) {
// additional test
{
sql: "select /*+ STREAM_AGG() */ distinct a from t",
best: "TableReader(Table(t)->StreamAgg)->StreamAgg",
best: "TableReader(Table(t))->StreamAgg",
},
{
sql: "select /*+ HASH_AGG() */ t1.a from t t1 where t1.a < any(select t2.b from t t2)",
Expand Down
36 changes: 25 additions & 11 deletions planner/core/task.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -873,15 +873,22 @@ func (p *PhysicalStreamAgg) attach2Task(tasks ...task) task {

// GetCost computes cost of stream aggregation considering CPU/memory.
func (p *PhysicalStreamAgg) GetCost(inputRows float64, isRoot bool) float64 {
numAggFunc := len(p.AggFuncs)
if numAggFunc == 0 {
numAggFunc = 1
aggFactorSum := 0.0
for _, agg := range p.AggFuncs {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why not extract a function for them?

if fac, ok := aggFuncFactor[agg.Name]; ok {
aggFactorSum += fac
} else {
aggFactorSum += aggFuncFactor["default"]
}
}
if aggFactorSum == 0 {
aggFactorSum = 1.0
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So no agg func has more cost that single bitXXX? That seems weird.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

My opinion is that bit operation costs lower than other base operations.

}
var cpuCost float64
if isRoot {
cpuCost = inputRows * cpuFactor * float64(numAggFunc)
cpuCost = inputRows * cpuFactor * aggFactorSum
} else {
cpuCost = inputRows * copCPUFactor * float64(numAggFunc)
cpuCost = inputRows * copCPUFactor * aggFactorSum
}
rowsPerGroup := inputRows / p.statsInfo().RowCount
memoryCost := rowsPerGroup * distinctFactor * memoryFactor * float64(p.numDistinctFunc())
Expand Down Expand Up @@ -954,23 +961,30 @@ func (p *PhysicalHashAgg) attach2Task(tasks ...task) task {
func (p *PhysicalHashAgg) GetCost(inputRows float64, isRoot bool) float64 {
cardinality := p.statsInfo().RowCount
numDistinctFunc := p.numDistinctFunc()
numAggFunc := len(p.AggFuncs)
if numAggFunc == 0 {
numAggFunc = 1
aggFactorSum := 0.0
for _, agg := range p.AggFuncs {
if fac, ok := aggFuncFactor[agg.Name]; ok {
aggFactorSum += fac
} else {
aggFactorSum += aggFuncFactor["default"]
}
}
if aggFactorSum == 0 {
aggFactorSum = 1.0
}
var cpuCost float64
if isRoot {
cpuCost = inputRows * cpuFactor * float64(numAggFunc)
cpuCost = inputRows * cpuFactor * aggFactorSum
divisor, con := p.cpuCostDivisor(numDistinctFunc > 0)
if divisor > 0 {
cpuCost /= divisor
// Cost of additional goroutines.
cpuCost += (con + 1) * concurrencyFactor
}
} else {
cpuCost = inputRows * copCPUFactor * float64(numAggFunc)
cpuCost = inputRows * copCPUFactor * aggFactorSum
}
memoryCost := cardinality * memoryFactor * float64(numAggFunc)
memoryCost := cardinality * memoryFactor * aggFactorSum
// When aggregation has distinct flag, we would allocate a map for each group to
// check duplication.
memoryCost += inputRows * distinctFactor * memoryFactor * float64(numDistinctFunc)
Expand Down