Changeset: ab10c261371e for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/ab10c261371e
Modified Files:
gdk/gdk_hash.c
gdk/gdk_private.h
gdk/gdk_string.c
Branch: ustr
Log Message:
Reduce the number of function calls made when inserting ustr data.
diffs (123 lines):
diff --git a/gdk/gdk_hash.c b/gdk/gdk_hash.c
--- a/gdk/gdk_hash.c
+++ b/gdk/gdk_hash.c
@@ -416,7 +416,8 @@ HASHgrowbucket(BAT *b)
HASHput(h, old, BUN_NONE);
else
HASHputlink(h, lold, BUN_NONE);
- } while (h->nunique >= (nbucket = h->nbucket) * 7 / 8);
+ nbucket = h->nbucket;
+ } while (h->nunique >= nbucket * 3 / 4);
TRC_DEBUG_IF(ACCELERATOR) if (h->nbucket > onbucket) {
TRC_DEBUG_ENDIF(ACCELERATOR, ALGOBATFMT " " BUNFMT
" -> " BUNFMT " buckets (" LLFMT " usec)\n",
@@ -436,7 +437,7 @@ HASHgrowbucket(BAT *b)
* maintained here, in the HASHdestroy and HASHfree functions, and in
* BBPdiskscan during initialization. */
bool
-BATcheckhash(BAT *b)
+BATcheckhash_locked(BAT *b)
{
if (b->ttype == TYPE_void)
return false;
@@ -444,13 +445,10 @@ BATcheckhash(BAT *b)
lng t = 0;
Hash *h;
- MT_rwlock_rdlock(&b->thashlock);
h = b->thash;
- MT_rwlock_rdunlock(&b->thashlock);
if (h == (Hash *) 1) {
/* but when we want to change it, we need the lock */
TRC_DEBUG_IF(ACCELERATOR) t = GDKusec();
- MT_rwlock_wrlock(&b->thashlock);
TRC_DEBUG_IF(ACCELERATOR) t = GDKusec() - t;
/* if still 1 now that we have the lock, we can update */
if (b->thash == (Hash *) 1) {
@@ -522,7 +520,6 @@ BATcheckhash(BAT *b)
h->heaplink.hasfile = true;
TRC_DEBUG(ACCELERATOR,
ALGOBATFMT ": reusing persisted hash\n", ALGOBATPAR(b));
-
MT_rwlock_wrunlock(&b->thashlock);
return true;
}
/* if h->nbucket
@@ -561,7 +558,6 @@ BATcheckhash(BAT *b)
GDKclrerr(); /* we're not currently interested in
errors */
}
h = b->thash;
- MT_rwlock_wrunlock(&b->thashlock);
}
if (h != NULL) {
TRC_DEBUG(ACCELERATOR, ALGOBATFMT ": already has hash, waited "
LLFMT " usec\n", ALGOBATPAR(b), t);
@@ -569,6 +565,25 @@ BATcheckhash(BAT *b)
return h != NULL;
}
+bool
+BATcheckhash(BAT *b)
+{
+ if (b->ttype == TYPE_void)
+ return false;
+
+ MT_rwlock_rdlock(&b->thashlock);
+ Hash *h = b->thash;
+ MT_rwlock_rdunlock(&b->thashlock);
+ if (h == NULL)
+ return false;
+ if (h != (Hash *) 1)
+ return true;
+ MT_rwlock_wrlock(&b->thashlock);
+ bool ret = BATcheckhash_locked(b);
+ MT_rwlock_wrunlock(&b->thashlock);
+ return ret;
+}
+
/* figure out size of the hash (sum of the sizes of the two hash files)
* without loading them */
size_t
@@ -1182,12 +1197,14 @@ HASHappend_locked_hashval(BATiter *bi, B
GDKclrerr();
return;
}
- if (HASHwidth(i + 1) > h->width &&
- HASHupgradehashheap(b) != GDK_SUCCEED) {
+ if (h->width < SIZEOF_BUN &&
+ HASHwidth(i + 1) > h->width &&
+ HASHupgradehashheap(b) != GDK_SUCCEED) {
GDKclrerr();
return;
}
if ((ATOMsize(b->ttype) > 2 &&
+ h->nunique >= h->nbucket * 7 / 8 &&
HASHgrowbucket(b) != GDK_SUCCEED) ||
((i + 1) * h->width > h->heaplink.size &&
HEAPextend(&h->heaplink,
diff --git a/gdk/gdk_private.h b/gdk/gdk_private.h
--- a/gdk/gdk_private.h
+++ b/gdk/gdk_private.h
@@ -45,6 +45,8 @@ void ATOMunknown_clean(void)
__attribute__((__visibility__("hidden")));
bool BATcheckhash(BAT *b)
__attribute__((__visibility__("hidden")));
+bool BATcheckhash_locked(BAT *b)
+ __attribute__((__visibility__("hidden")));
gdk_return BATcheckmodes(BAT *b, bool persistent)
__attribute__((__warn_unused_result__))
__attribute__((__visibility__("hidden")));
diff --git a/gdk/gdk_string.c b/gdk/gdk_string.c
--- a/gdk/gdk_string.c
+++ b/gdk/gdk_string.c
@@ -185,10 +185,11 @@ ustrPut(BAT *b, var_t *dst, const char *
/* this function is the ONLY place where ustrbat may be changed
* (inserted into), and we're holding the ustrbat->theaplock, so we are
* totally save looking at the heaps */
- (void) BATcheckhash(ustrbat); /* load hash table */
MT_rwlock_wrlock(&ustrbat->thashlock);
MT_lock_set(&ustrbat->theaplock);
assert(ustrbat->tvkey);
+ if (b->thash == (Hash *) 1)
+ (void) BATcheckhash_locked(ustrbat); /* load hash table */
Heap *h = ustrbat->tvheap;
BUN p = BUN_NONE;
_______________________________________________
checkin-list mailing list -- [email protected]
To unsubscribe send an email to [email protected]
