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

statistics: reduce memory usage when to MergePartTopN2GlobalTopN #45718

Merged
merged 3 commits into from
Aug 10, 2023
Merged

statistics: reduce memory usage when to MergePartTopN2GlobalTopN #45718

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
387d101
statistics: reduce memory usage when to MergePartTopN2GlobalTopN
hawkingrei Aug 7, 2023
715b5d2
statistics: reduce memory usage when to MergePartTopN2GlobalTopN
hawkingrei Aug 7, 2023
8ff431d
update
hawkingrei Aug 8, 2023
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
16 changes: 1 addition & 15 deletions statistics/cmsketch.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -799,10 +799,7 @@ func MergePartTopN2GlobalTopN(loc *time.Location, version int, topNs []*TopN, n
if checkEmptyTopNs(topNs) {
return nil, nil, hists, nil
}

partNum := len(topNs)
removeVals := make([][]TopNMeta, partNum)

// Different TopN structures may hold the same value, we have to merge them.
counter := make(map[hack.MutableString]float64)
// datumMap is used to store the mapping from the string type to datum type.
Expand Down Expand Up @@ -864,22 +861,11 @@ func MergePartTopN2GlobalTopN(loc *time.Location, version int, topNs []*TopN, n
if count != 0 {
counter[encodedVal] += count
// Remove the value corresponding to encodedVal from the histogram.
removeVals[j] = append(removeVals[j], TopNMeta{Encoded: datum.GetBytes(), Count: uint64(count)})
hists[j].BinarySearchRemoveVal(TopNMeta{Encoded: datum.GetBytes(), Count: uint64(count)})
}
}
}
}
// Remove the value from the Hists.
for i := 0; i < partNum; i++ {
if len(removeVals[i]) > 0 {
tmp := removeVals[i]
slices.SortFunc(tmp, func(i, j TopNMeta) bool {
cmpResult := bytes.Compare(i.Encoded, j.Encoded)
return cmpResult < 0
})
hists[i].RemoveVals(tmp)
}
}
numTop := len(counter)
if numTop == 0 {
return nil, nil, hists, nil
Expand Down
29 changes: 29 additions & 0 deletions statistics/histogram.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -281,6 +281,35 @@ func (hg *Histogram) BucketToString(bktID, idxCols int) string {
return fmt.Sprintf("num: %d lower_bound: %s upper_bound: %s repeats: %d ndv: %d", hg.bucketCount(bktID), lowerVal, upperVal, hg.Buckets[bktID].Repeat, hg.Buckets[bktID].NDV)
}

// BinarySearchRemoveVal removes the value from the TopN using binary search.
func (hg *Histogram) BinarySearchRemoveVal(valCntPairs TopNMeta) {
Copy link
Member Author

Choose a reason for hiding this comment

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

BinarySearchRemoveVal is the same as the RemoveVal.

Copy link
Member

Choose a reason for hiding this comment

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

There have been several other methods to find a value in the Histogram, it may be better to try to reuse them, like hg.locateBucket() and hg.Bounds.LowerBound().

Copy link
Member Author

Choose a reason for hiding this comment

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

I find this problem too. but because this PR need to cherry-pick for old PR. we need to make changement simple. I will improve this code next next pr.

lowIdx, highIdx := 0, hg.Len()-1
for lowIdx <= highIdx {
midIdx := (lowIdx + highIdx) / 2
cmpResult := bytes.Compare(hg.Bounds.Column(0).GetRaw(midIdx*2), valCntPairs.Encoded)
if cmpResult > 0 {
lowIdx = midIdx + 1
continue
}
cmpResult = bytes.Compare(hg.Bounds.Column(0).GetRaw(midIdx*2+1), valCntPairs.Encoded)
if cmpResult < 0 {
highIdx = midIdx - 1
continue
}
if hg.Buckets[midIdx].NDV > 0 {
hg.Buckets[midIdx].NDV--
}
if cmpResult == 0 {
hg.Buckets[midIdx].Repeat = 0
}
hg.Buckets[midIdx].Count -= int64(valCntPairs.Count)
if hg.Buckets[midIdx].Count < 0 {
hg.Buckets[midIdx].Count = 0
}
break
}
}

// RemoveVals remove the given values from the histogram.
// This function contains an **ASSUMPTION**: valCntPairs is sorted in ascending order.
func (hg *Histogram) RemoveVals(valCntPairs []TopNMeta) {
Expand Down