Changeset: a720809a0802 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/a720809a0802
Modified Files:
gdk/gdk_group.c
Branch: dict
Log Message:
Implemented special BATgroup case for byte-sized values with limited number of
pre-existing groups.
diffs (127 lines):
diff --git a/gdk/gdk_group.c b/gdk/gdk_group.c
--- a/gdk/gdk_group.c
+++ b/gdk/gdk_group.c
@@ -691,9 +691,8 @@ BATgroup_internal(BAT **groups, BAT **ex
}
assert(!BATtdense(b));
if (g) {
- if (BATtdense(g))
- maxgrp = g->tseqbase + BATcount(g);
- else if (BATtordered(g))
+ assert(!BATtdense(g));
+ if (BATtordered(g))
maxgrp = * (oid *) Tloc(g, BATcount(g) - 1);
else if (BATtrevordered(g))
maxgrp = * (oid *) Tloc(g, 0);
@@ -965,10 +964,10 @@ BATgroup_internal(BAT **groups, BAT **ex
/* byte-sized values, use 256 entry array to keep
* track of doled out group ids; note that we can't
* possibly have more than 256 groups, so the group id
- * fits in an unsigned char */
- unsigned char *restrict bgrps = GDKmalloc(256);
- const unsigned char *restrict w = (const unsigned char *)
bi.base;
- unsigned char v;
+ * fits in a uint8_t */
+ uint8_t *restrict bgrps = GDKmalloc(256);
+ const uint8_t *restrict w = (const uint8_t *) bi.base;
+ uint8_t v;
algomsg = "byte-sized groups -- ";
if (bgrps == NULL)
@@ -982,7 +981,7 @@ BATgroup_internal(BAT **groups, BAT **ex
oid o = canditer_next(&ci);
p = o - b->hseqbase;
if ((v = bgrps[w[p]]) == 0xFF && ngrp < 256) {
- bgrps[w[p]] = v = (unsigned char) ngrp++;
+ bgrps[w[p]] = v = (uint8_t) ngrp++;
maxgrppos = r;
if (extents)
exts[v] = o;
@@ -996,19 +995,55 @@ BATgroup_internal(BAT **groups, BAT **ex
TIMEOUT_CHECK(timeoffset,
GOTO_LABEL_TIMEOUT_HANDLER(error));
GDKfree(bgrps);
+ } else if (t == TYPE_bte && maxgrp < 256) {
+ /* byte-sized values with a limited number of groups,
+ * use 65536 entry array to keep track of doled out
+ * group ids; note that we can't possibly have more than
+ * 65536 goups, so the group id fits in a uint16_t */
+ uint16_t *restrict sgrps = GDKmalloc(65536 * sizeof(uint16_t));
+ const uint8_t *restrict w = (const uint8_t *) bi.base;
+ uint16_t v;
+
+ algomsg = "short-sized subgroups -- ";
+ if (sgrps == NULL)
+ goto error1;
+ memset(sgrps, 0xFF, 65536 * sizeof(uint16_t));
+ if (histo)
+ memset(cnts, 0, maxgrps * sizeof(lng));
+ ngrp = 0;
+ gn->tsorted = true;
+ TIMEOUT_LOOP_IDX(r, cnt, timeoffset) {
+ oid o = canditer_next(&ci);
+ p = o - b->hseqbase;
+ uint16_t x = (uint16_t) (w[p] | (grps[p] << 8));
+ if ((v = sgrps[x]) == 0xFFFF && ngrp < 65536) {
+ sgrps[x] = v = (uint16_t) ngrp++;
+ maxgrppos = r;
+ if (extents)
+ exts[v] = o;
+ }
+ ngrps[r] = v;
+ if (r > 0 && v < ngrps[r - 1])
+ gn->tsorted = false;
+ if (histo)
+ cnts[v]++;
+ }
+ TIMEOUT_CHECK(timeoffset,
+ GOTO_LABEL_TIMEOUT_HANDLER(error));
+ GDKfree(sgrps);
} else if (g == NULL && t == TYPE_sht) {
/* short-sized values, use 65536 entry array to keep
* track of doled out group ids; note that we can't
* possibly have more than 65536 groups, so the group
- * id fits in an unsigned short */
- unsigned short *restrict sgrps = GDKmalloc(65536 *
sizeof(short));
- const unsigned short *restrict w = (const unsigned short *)
bi.base;
- unsigned short v;
+ * id fits in a uint16_t */
+ uint16_t *restrict sgrps = GDKmalloc(65536 * sizeof(uint16_t));
+ const uint16_t *restrict w = (const uint16_t *) bi.base;
+ uint16_t v;
algomsg = "short-sized groups -- ";
if (sgrps == NULL)
goto error1;
- memset(sgrps, 0xFF, 65536 * sizeof(short));
+ memset(sgrps, 0xFF, 65536 * sizeof(uint16_t));
if (histo)
memset(cnts, 0, maxgrps * sizeof(lng));
ngrp = 0;
@@ -1017,7 +1052,7 @@ BATgroup_internal(BAT **groups, BAT **ex
oid o = canditer_next(&ci);
p = o - b->hseqbase;
if ((v = sgrps[w[p]]) == 0xFFFF && ngrp < 65536) {
- sgrps[w[p]] = v = (unsigned short) ngrp++;
+ sgrps[w[p]] = v = (uint16_t) ngrp++;
maxgrppos = r;
if (extents)
exts[v] = o;
@@ -1190,7 +1225,7 @@ BATgroup_internal(BAT **groups, BAT **ex
const bte *w = (bte *) bi.base;
GRP_create_partial_hash_table_core(
(void) 0,
- (v = ((ulng)grps[r]<<8)|(unsigned
char)w[p], hash_lng(hs, &v)),
+ (v = ((ulng)grps[r]<<8)|(uint8_t)w[p],
hash_lng(hs, &v)),
w[p] == w[hb] && grps[r] == grps[q],
(void) 0,
NOGRPTST);
@@ -1207,7 +1242,7 @@ BATgroup_internal(BAT **groups, BAT **ex
const sht *w = (sht *) bi.base;
GRP_create_partial_hash_table_core(
(void) 0,
- (v = ((ulng)grps[r]<<16)|(unsigned
short)w[p], hash_lng(hs, &v)),
+ (v =
((ulng)grps[r]<<16)|(uint16_t)w[p], hash_lng(hs, &v)),
w[p] == w[hb] && grps[r] == grps[q],
(void) 0,
NOGRPTST);
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
