Changeset: b3341f69aab6 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=b3341f69aab6
Modified Files:
gdk/gdk_join.c
Branch: Oct2014
Log Message:
Fix the previous fix: fewer extends by better extrapolation.
diffs (39 lines):
diff --git a/gdk/gdk_join.c b/gdk/gdk_join.c
--- a/gdk/gdk_join.c
+++ b/gdk/gdk_join.c
@@ -350,6 +350,7 @@ mergejoin(BAT *r1, BAT *r2, BAT *l, BAT
int (*cmp)(const void *, const void *) = BATatoms[l->ttype].atomCmp;
const char *v, *prev = NULL;
BUN nl, nr;
+ BUN total; /* number of rows in l we scan */
int insert_nil;
/* equal_order is set if we can scan both BATs in the same
* order, so when both are sorted or both are reverse sorted
@@ -394,6 +395,7 @@ mergejoin(BAT *r1, BAT *r2, BAT *l, BAT
CANDINIT(l, sl, lstart, lend, lcnt, lcand, lcandend);
CANDINIT(r, sr, rstart, rend, rcnt, rcand, rcandend);
+ total = lcand ? (BUN) (lcandend - lcand) : lend - lstart;
lvals = l->ttype == TYPE_void ? NULL : (const char *) Tloc(l,
BUNfirst(l));
rvals = r->ttype == TYPE_void ? NULL : (const char *) Tloc(r,
BUNfirst(r));
if (l->tvarsized && l->ttype) {
@@ -1050,12 +1052,15 @@ mergejoin(BAT *r1, BAT *r2, BAT *l, BAT
* we need to add nl * nr values in the results */
if (BATcount(r1) + nl * nr > BATcapacity(r1)) {
/* make some extra space by extrapolating how
- * much more we need (used part of l divided
- * by total length of l, multiplied by amount
- * generated so far) */
- BUN newcap = (BUN) ((lcand ? (double) (lcandend + 1 -
lcand) / BATcount(sl) : (double) (lend + 1 - lstart) / lend) * (BATcount(r1) +
nl * nr));
+ * much more we need (fraction of l we've seen
+ * so far is used as the fraction of the
+ * expected result size we've produced so
+ * far) */
+ BUN newcap = (BUN) ((double) total / (total - (lcand ?
(BUN) (lcandend - lcand) : (lend - lstart))) * (BATcount(r1) + nl * nr) * 1.1);
if (newcap < nl * nr + BATcount(r1))
newcap = nl * nr + BATcount(r1) + 1024;
+ /* make sure heap.free is set properly before
+ * extending */
BATsetcount(r1, BATcount(r1));
BATsetcount(r2, BATcount(r2));
r1 = BATextend(r1, newcap);
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
