Changeset: bb0575cfe1f2 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=bb0575cfe1f2
Modified Files:
gdk/gdk.h
gdk/gdk_batop.c
gdk/gdk_join.c
gdk/gdk_project.c
Branch: Oct2020
Log Message:
Merged with Jun2020
diffs (truncated from 360 to 300 lines):
diff --git a/gdk/gdk.h b/gdk/gdk.h
--- a/gdk/gdk.h
+++ b/gdk/gdk.h
@@ -2023,6 +2023,7 @@ enum prop_t {
GDK_MAX_VALUE, /* largest non-nil value in BAT */
GDK_HASH_BUCKETS, /* last used hash bucket size */
GDK_NUNIQUE, /* number of unique values */
+ GDK_UNIQUE_ESTIMATE, /* estimate of number of distinct values */
};
/*
diff --git a/gdk/gdk_bat.c b/gdk/gdk_bat.c
--- a/gdk/gdk_bat.c
+++ b/gdk/gdk_bat.c
@@ -1073,6 +1073,7 @@ BUNappend(BAT *b, const void *t, bool fo
IMPSdestroy(b); /* no support for inserts in imprints yet */
OIDXdestroy(b);
BATrmprop(b, GDK_NUNIQUE);
+ BATrmprop(b, GDK_UNIQUE_ESTIMATE);
#if 0 /* enable if we have more properties than just min/max */
PROPrec *prop;
do {
@@ -1159,6 +1160,7 @@ BUNdelete(BAT *b, oid o)
OIDXdestroy(b);
HASHdestroy(b);
BATrmprop(b, GDK_NUNIQUE);
+ BATrmprop(b, GDK_UNIQUE_ESTIMATE);
#if 0 /* enable if we have more properties than just min/max */
do {
for (prop = b->tprops; prop; prop = prop->next)
@@ -1248,6 +1250,7 @@ BUNinplace(BAT *b, BUN p, const void *t,
}
}
BATrmprop(b, GDK_NUNIQUE);
+ BATrmprop(b, GDK_UNIQUE_ESTIMATE);
#if 0 /* enable if we have more properties than just min/max */
do {
for (prop = b->tprops; prop; prop = prop->next)
diff --git a/gdk/gdk_batop.c b/gdk/gdk_batop.c
--- a/gdk/gdk_batop.c
+++ b/gdk/gdk_batop.c
@@ -573,6 +573,7 @@ BATappend2(BAT *b, BAT *n, BAT *s, bool
}
}
BATrmprop(b, GDK_NUNIQUE);
+ BATrmprop(b, GDK_UNIQUE_ESTIMATE);
#if 0 /* enable if we have more properties than just min/max */
do {
for (prop = b->tprops; prop; prop = prop->next)
@@ -890,6 +891,7 @@ BATreplace(BAT *b, BAT *p, BAT *n, bool
OIDXdestroy(b);
IMPSdestroy(b);
BATrmprop(b, GDK_NUNIQUE);
+ BATrmprop(b, GDK_UNIQUE_ESTIMATE);
b->tsorted = b->trevsorted = false;
b->tnosorted = b->tnorevsorted = 0;
diff --git a/gdk/gdk_join.c b/gdk/gdk_join.c
--- a/gdk/gdk_join.c
+++ b/gdk/gdk_join.c
@@ -3041,12 +3041,19 @@ guess_uniques(BAT *b, struct canditer *c
if (b->tkey)
return (double) ci->ncand;
- if (ci->s) {
+ if (ci->s == NULL ||
+ (ci->tpe == cand_dense && ci->ncand == BATcount(b))) {
+ PROPrec *p = BATgetprop(b, GDK_UNIQUE_ESTIMATE);
+ if (p) {
+ TRC_DEBUG(ALGO, "b=" ALGOBATFMT " use cached value\n",
+ ALGOBATPAR(b));
+ return p->v.val.dval;
+ }
+ s1 = BATsample(b, 1000);
+ } else {
BAT *s2 = BATsample(ci->s, 1000);
s1 = BATproject(s2, ci->s);
BBPreclaim(s2);
- } else {
- s1 = BATsample(b, 1000);
}
BUN n2 = BATcount(s1);
BUN n1 = n2 / 2;
@@ -3056,7 +3063,12 @@ guess_uniques(BAT *b, struct canditer *c
double A = (double) (cnt2 - cnt1) / (n2 - n1);
double B = cnt1 - n1 * A;
- return A * ci->ncand + B;
+ B += A * ci->ncand;
+ if (ci->s == NULL ||
+ (ci->tpe == cand_dense && ci->ncand == BATcount(b))) {
+ BATsetprop(b, GDK_UNIQUE_ESTIMATE, TYPE_dbl, &B);
+ }
+ return B;
}
#define MASK_EQ 1
diff --git a/gdk/gdk_project.c b/gdk/gdk_project.c
--- a/gdk/gdk_project.c
+++ b/gdk/gdk_project.c
@@ -290,6 +290,211 @@ project_any(BAT *restrict bn, BAT *restr
return GDK_SUCCEED;
}
+static BAT *
+project_str(BAT *restrict l, struct canditer *restrict ci,
+ BAT *restrict r1, BAT *restrict r2, lng t0)
+{
+ BAT *bn;
+ BUN lo, hi;
+ oid r1seq, r1end;
+ oid r2seq, r2end;
+ BUN h1off;
+ BAT *r;
+ BUN off;
+ oid seq;
+ var_t v;
+
+ if ((bn = COLnew(l->hseqbase, TYPE_str, ci ? ci->ncand : BATcount(l),
+ TRANSIENT)) != NULL)
+ return NULL;
+
+ v = (var_t) r1->tvheap->free;
+ if (r1->tvheap == r2->tvheap) {
+ h1off = 0;
+ BBPshare(bn->tvheap->parentid);
+ HEAPfree(bn->tvheap, true);
+ GDKfree(bn->tvheap);
+ bn->tvheap = r1->tvheap;
+ } else {
+ v = (v + GDK_VARALIGN - 1) & ~(GDK_VARALIGN - 1);
+ h1off = (BUN) v;
+ v += ((var_t) r2->tvheap->free + GDK_VARALIGN - 1) &
~(GDK_VARALIGN - 1);
+ if (HEAPextend(bn->tvheap, v, false) != GDK_SUCCEED) {
+ BBPreclaim(bn);
+ return NULL;
+ }
+ memcpy(bn->tvheap->base, r1->tvheap->base, r1->tvheap->free);
+#ifndef NDEBUG
+ if (h1off > r1->tvheap->free)
+ memset(bn->tvheap->base + r1->tvheap->free, 0, h1off -
r1->tvheap->free);
+#endif
+ memcpy(bn->tvheap->base + h1off, r2->tvheap->base,
r2->tvheap->free);
+ }
+
+ if (v >= ((var_t) 1 << (8 * bn->twidth)) &&
+ GDKupgradevarheap(bn, v, false, false) != GDK_SUCCEED) {
+ BBPreclaim(bn);
+ return NULL;
+ }
+
+ r1seq = r1->hseqbase;
+ r1end = r1seq + BATcount(r1);
+ r2seq = r2->hseqbase;
+ r2end = r2seq + BATcount(r2);
+ if (ci) {
+ for (lo = 0, hi = ci->ncand; lo < hi; lo++) {
+ oid o = canditer_next(ci);
+ if (o < r1seq || o >= r2end) {
+ GDKerror("does not match always\n");
+ return GDK_FAIL;
+ }
+ if (o < r1end) {
+ r = r1;
+ off = 0;
+ seq = r1seq;
+ } else {
+ r = r2;
+ off = h1off;
+ seq = r2seq;
+ }
+ switch (r->twidth) {
+ case 1:
+ v = (var_t) ((uint8_t *) r->theap.base)[o -
seq] + GDK_VAROFFSET;
+ break;
+ case 2:
+ v = (var_t) ((uint16_t *) r->theap.base)[o -
seq] + GDK_VAROFFSET;
+ break;
+ case 4:
+ v = (var_t) ((uint32_t *) r->theap.base)[o -
seq];
+ break;
+ case 8:
+ v = (var_t) ((uint64_t *) r->theap.base)[o -
seq];
+ break;
+ }
+ v += off;
+ switch (bn->twidth) {
+ case 1:
+ ((uint8_t *) bn->theap.base)[lo] = v -
GDK_VAROFFSET;
+ break;
+ case 2:
+ ((uint16_t *) bn->theap.base)[lo] = v -
GDK_VAROFFSET;
+ break;
+ case 4:
+ ((uint32_t *) bn->theap.base)[lo] = (uint32_t)
v;
+ break;
+ case 8:
+ ((uint64_t *) bn->theap.base)[lo] = (uint64_t)
v;
+ break;
+ }
+ }
+ } else if (BATtdense(l)) {
+ for (lo = 0, hi = BATcount(l); lo < hi; lo++) {
+ oid o = l->tseqbase + lo;
+ if (o < r1seq || o >= r2end) {
+ GDKerror("does not match always\n");
+ return GDK_FAIL;
+ }
+ if (o < r1end) {
+ r = r1;
+ off = 0;
+ seq = r1seq;
+ } else {
+ r = r2;
+ off = h1off;
+ seq = r2seq;
+ }
+ switch (r->twidth) {
+ case 1:
+ v = (var_t) ((uint8_t *) r->theap.base)[o -
seq] + GDK_VAROFFSET;
+ break;
+ case 2:
+ v = (var_t) ((uint16_t *) r->theap.base)[o -
seq] + GDK_VAROFFSET;
+ break;
+ case 4:
+ v = (var_t) ((uint32_t *) r->theap.base)[o -
seq];
+ break;
+ case 8:
+ v = (var_t) ((uint64_t *) r->theap.base)[o -
seq];
+ break;
+ }
+ v += off;
+ switch (bn->twidth) {
+ case 1:
+ ((uint8_t *) bn->theap.base)[lo] = v -
GDK_VAROFFSET;
+ break;
+ case 2:
+ ((uint16_t *) bn->theap.base)[lo] = v -
GDK_VAROFFSET;
+ break;
+ case 4:
+ ((uint32_t *) bn->theap.base)[lo] = (uint32_t)
v;
+ break;
+ case 8:
+ ((uint64_t *) bn->theap.base)[lo] = (uint64_t)
v;
+ break;
+ }
+ }
+ } else {
+ const oid *restrict ot = (const oid *) Tloc(l, 0);
+ for (lo = 0, hi = BATcount(l); lo < hi; lo++) {
+ oid o = ot[lo];
+ if (o < r1seq || o >= r2end) {
+ GDKerror("does not match always\n");
+ return GDK_FAIL;
+ }
+ if (o < r1end) {
+ r = r1;
+ off = 0;
+ seq = r1seq;
+ } else {
+ r = r2;
+ off = h1off;
+ seq = r2seq;
+ }
+ switch (r->twidth) {
+ case 1:
+ v = (var_t) ((uint8_t *) r->theap.base)[o -
seq] + GDK_VAROFFSET;
+ break;
+ case 2:
+ v = (var_t) ((uint16_t *) r->theap.base)[o -
seq] + GDK_VAROFFSET;
+ break;
+ case 4:
+ v = (var_t) ((uint32_t *) r->theap.base)[o -
seq];
+ break;
+ case 8:
+ v = (var_t) ((uint64_t *) r->theap.base)[o -
seq];
+ break;
+ }
+ v += off;
+ switch (bn->twidth) {
+ case 1:
+ ((uint8_t *) bn->theap.base)[lo] = v -
GDK_VAROFFSET;
+ break;
+ case 2:
+ ((uint16_t *) bn->theap.base)[lo] = v -
GDK_VAROFFSET;
+ break;
+ case 4:
+ ((uint32_t *) bn->theap.base)[lo] = (uint32_t)
v;
+ break;
+ case 8:
+ ((uint64_t *) bn->theap.base)[lo] = (uint64_t)
v;
+ break;
+ }
+ }
+ }
+ BATsetcount(bn, lo);
+ bn->tsorted = bn->trevsorted = false;
+ bn->tnil = false;
+ bn->tnonil = r1->tnonil & r2->tnonil;
+ bn->tkey = false;
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
