cp to 1.2-dev 'fix ndv calculation to make it more accurate' (#18848)

fix ndv calculation to make it more accurate Approved by: @triump2020, @XuPeng-SH, @sukki37
matrixorigin · Sep 19, 2024 · 5cd387f · 5cd387f
1 parent 75ae7b4
commit 5cd387f
Showing 1 changed file with 6 additions and 0 deletions.
diff --git a/pkg/vm/engine/disttae/stats.go b/pkg/vm/engine/disttae/stats.go
@@ -783,6 +783,12 @@ func adjustNDV(info *plan2.InfoFromZoneMap, tableDef *plan2.TableDef) {
 			}
 			if rate < 0.1 {
 				info.ColumnNDVs[idx] /= math.Pow(float64(info.AccurateObjectNumber), (1 - rate))
+				if info.ColumnNDVs[idx] > 500 {
+					info.ColumnNDVs[idx] *= math.Pow(info.ColumnNDVs[idx], 0.2)
+					if info.ColumnSize[idx] > 0 {
+						info.ColumnNDVs[idx] *= math.Pow(float64(info.ColumnSize[idx]), 0.2)
+					}
+				}
 			}
 			ndvUsingZonemap := calcNdvUsingZonemap(info.ColumnZMs[idx], &info.DataTypes[idx])
 			if ndvUsingZonemap != -1 && info.ColumnNDVs[idx] > ndvUsingZonemap {