Changeset: ed3714394359 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/ed3714394359
Modified Files:
gdk/gdk_imprints.c
gdk/gdk_join.c
gdk/gdk_private.h
gdk/gdk_sample.c
gdk/xoshiro256starstar.h
sql/test/sample/Tests/sample.10.test
sql/test/sample/Tests/sample.11.test
sql/test/sample/Tests/sample.12.test
sql/test/sample/Tests/sample.13.test
sql/test/sample/Tests/sample.2.test
sql/test/sample/Tests/sample.3.test
sql/test/sample/Tests/sample.4.test
sql/test/sample/Tests/sample.5.test
sql/test/sample/Tests/sample.6.test
sql/test/sample/Tests/sample.9.test
sql/test/sys-schema/Tests/webExamplesMathematicalFunctionsOperators.test
Branch: Dec2023
Log Message:
Updated BATsample, remove BATcreatesample (which was private anyway).
diffs (truncated from 579 to 300 lines):
diff --git a/gdk/gdk_imprints.c b/gdk/gdk_imprints.c
--- a/gdk/gdk_imprints.c
+++ b/gdk/gdk_imprints.c
@@ -523,7 +523,7 @@ BATimprints(BAT *b)
imprints->imprints.parentid = b->batCacheid;
#define SMP_SIZE 2048
- s1 = BATsample_with_seed(b, SMP_SIZE, (uint64_t) GDKusec() *
(uint64_t) b->batCacheid);
+ s1 = BATsample(b, SMP_SIZE);
if (s1 == NULL) {
GDKfree(imprints);
bat_iterator_end(&bi);
diff --git a/gdk/gdk_join.c b/gdk/gdk_join.c
--- a/gdk/gdk_join.c
+++ b/gdk/gdk_join.c
@@ -3559,11 +3559,9 @@ guess_uniques(BAT *b, struct canditer *c
ALGOBATPAR(b));
return unique_est;
}
- s1 = BATcreatesample(b->hseqbase, batcount, 1000,
- (uint64_t) GDKusec() * (uint64_t)
b->batCacheid);
+ s1 = BATsample(b, 1000);
} else {
- BAT *s2 = BATcreatesample(ci->s->hseqbase, ci->ncand, 1000,
- (uint64_t) GDKusec() * (uint64_t)
b->batCacheid);
+ BAT *s2 = BATsample(ci->s, 1000);
if (s2 == NULL)
return -1;
s1 = BATproject(s2, ci->s);
diff --git a/gdk/gdk_private.h b/gdk/gdk_private.h
--- a/gdk/gdk_private.h
+++ b/gdk/gdk_private.h
@@ -67,8 +67,6 @@ gdk_return BATcheckmodes(BAT *b, bool pe
__attribute__((__visibility__("hidden")));
BAT *BATcreatedesc(oid hseq, int tt, bool heapnames, role_t role, uint16_t
width)
__attribute__((__visibility__("hidden")));
-BAT *BATcreatesample(oid hseq, BUN cnt, BUN n, uint64_t seed)
- __attribute__((__visibility__("hidden")));
void BATdelete(BAT *b)
__attribute__((__visibility__("hidden")));
void BATdestroy(BAT *b)
diff --git a/gdk/gdk_sample.c b/gdk/gdk_sample.c
--- a/gdk/gdk_sample.c
+++ b/gdk/gdk_sample.c
@@ -42,7 +42,7 @@ struct oidtreenode {
};
};
-static int
+static bool
OIDTreeMaybeInsert(struct oidtreenode *tree, oid o, BUN allocated)
{
struct oidtreenode **nodep;
@@ -50,12 +50,12 @@ OIDTreeMaybeInsert(struct oidtreenode *t
if (allocated == 0) {
tree->left = tree->right = NULL;
tree->o = o;
- return 1;
+ return true;
}
nodep = &tree;
while (*nodep) {
if (o == (*nodep)->o)
- return 0;
+ return false;
if (o < (*nodep)->o)
nodep = &(*nodep)->left;
else
@@ -64,7 +64,7 @@ OIDTreeMaybeInsert(struct oidtreenode *t
*nodep = &tree[allocated];
tree[allocated].left = tree[allocated].right = NULL;
tree[allocated].o = o;
- return 1;
+ return true;
}
/* inorder traversal, gives us a sorted BAT */
@@ -134,19 +134,14 @@ do_batsample(oid hseq, BUN cnt, BUN n, r
return NULL;
}
+ if (lock)
+ MT_lock_set(lock);
/* generate a list of random numbers; note we use the
* "tree" array, but we use the value from each location
* before it is overwritten by the use as part of the
* binary tree */
- if (lock) {
- MT_lock_set(lock);
- for (rescnt = 0; rescnt < n; rescnt++)
- tree[rescnt].r = next(rse);
- MT_lock_unset(lock);
- } else {
- for (rescnt = 0; rescnt < n; rescnt++)
- tree[rescnt].r = next(rse);
- }
+ for (rescnt = 0; rescnt < n; rescnt++)
+ tree[rescnt].r = next(rse);
/* while we do not have enough sample OIDs yet */
BUN rnd = 0;
@@ -156,12 +151,8 @@ do_batsample(oid hseq, BUN cnt, BUN n, r
if (rnd == n) {
/* we ran out of random numbers,
* so generate more */
- if (lock)
- MT_lock_set(lock);
for (rnd = rescnt; rnd < n; rnd++)
tree[rnd].r = next(rse);
- if (lock)
- MT_lock_unset(lock);
rnd = rescnt;
}
candoid = minoid + tree[rnd++].r % cnt;
@@ -169,6 +160,8 @@ do_batsample(oid hseq, BUN cnt, BUN n, r
* generated, try again */
} while (!OIDTreeMaybeInsert(tree, candoid, rescnt));
}
+ if (lock)
+ MT_lock_unset(lock);
if (!antiset) {
OIDTreeToBAT(tree, bn);
} else {
@@ -185,26 +178,20 @@ do_batsample(oid hseq, BUN cnt, BUN n, r
return bn;
}
+/* BATsample implements sampling for BATs */
BAT *
-BATcreatesample(oid hseq, BUN cnt, BUN n, uint64_t seed)
+BATsample_with_seed(BAT *b, BUN n, uint64_t seed)
{
random_state_engine rse;
init_random_state_engine(rse, seed);
- BAT *bn = do_batsample(hseq, cnt, n, rse, NULL);
- TRC_DEBUG(ALGO, OIDFMT "," BUNFMT "," BUNFMT " -> " ALGOOPTBATFMT "\n",
- hseq, cnt, n, ALGOOPTBATPAR(bn));
+ BAT *bn = do_batsample(b->hseqbase, BATcount(b), n, rse, NULL);
+ TRC_DEBUG(ALGO, ALGOBATFMT "," BUNFMT " -> " ALGOOPTBATFMT "\n",
+ ALGOBATPAR(b), n, ALGOOPTBATPAR(bn));
return bn;
}
-/* BATsample implements sampling for BATs */
-BAT *
-BATsample_with_seed(BAT *b, BUN n, uint64_t seed)
-{
- return BATcreatesample(b->hseqbase, b->batCount, n, seed);
-}
-
static MT_Lock rse_lock = MT_LOCK_INITIALIZER(rse_lock);
BAT *
BATsample(BAT *b, BUN n)
diff --git a/gdk/xoshiro256starstar.h b/gdk/xoshiro256starstar.h
--- a/gdk/xoshiro256starstar.h
+++ b/gdk/xoshiro256starstar.h
@@ -17,12 +17,14 @@ worldwide. This software is distributed
See <http://creativecommons.org/publicdomain/zero/1.0/>. */
-static inline uint64_t rotl(const uint64_t x, int k) {
+typedef uint64_t random_state_engine[4];
+
+static inline uint64_t
+rotl(const uint64_t x, int k)
+{
return (x << k) | (x >> (64 - k));
}
-typedef uint64_t random_state_engine[4];
-
static inline void
init_random_state_engine(random_state_engine engine, uint64_t seed)
{
@@ -40,8 +42,10 @@ init_random_state_engine(random_state_en
}
}
-static inline uint64_t next(random_state_engine rse) {
- const uint64_t output = rotl(rse[0] * 5, 7) * 9;
+static inline uint64_t
+next(random_state_engine rse)
+{
+ const uint64_t output = rotl(rse[1] * 5, 7) * 9;
const uint64_t t = rse[1] << 17;
diff --git a/sql/test/sample/Tests/sample.10.test
b/sql/test/sample/Tests/sample.10.test
--- a/sql/test/sample/Tests/sample.10.test
+++ b/sql/test/sample/Tests/sample.10.test
@@ -1,14 +1,14 @@
query I rowsort
select * from GENERATE_SERIES(1,41) sample 0.25 seed 1234
----
-14
-15
-19
-2
+1
+16
+17
20
+22
26
-27
-38
-39
-8
+28
+29
+31
+7
diff --git a/sql/test/sample/Tests/sample.11.test
b/sql/test/sample/Tests/sample.11.test
--- a/sql/test/sample/Tests/sample.11.test
+++ b/sql/test/sample/Tests/sample.11.test
@@ -1,24 +1,24 @@
query I rowsort
select * from GENERATE_SERIES(1,41) sample 0.5 seed 1234
----
-12
-13
-14
-15
+1
+11
+16
17
18
-19
2
20
+22
26
27
+28
+29
3
30
-37
-38
-39
-4
+31
+35
+36
+5
7
8
-9
diff --git a/sql/test/sample/Tests/sample.12.test
b/sql/test/sample/Tests/sample.12.test
--- a/sql/test/sample/Tests/sample.12.test
+++ b/sql/test/sample/Tests/sample.12.test
@@ -1,34 +1,34 @@
query I rowsort
select * from GENERATE_SERIES(1,41) sample 0.75 seed 1234
----
-1
10
11
12
13
-16
-17
+14
+15
18
+19
+2
21
-22
23
24
25
-28
-29
+27
3
30
-31
32
33
34
35
36
37
+38
+39
4
40
5
6
-7
+8
9
diff --git a/sql/test/sample/Tests/sample.13.test
b/sql/test/sample/Tests/sample.13.test
_______________________________________________
checkin-list mailing list -- [email protected]
To unsubscribe send an email to [email protected]
