Changeset: cb334b0ff7d0 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/cb334b0ff7d0
Modified Files:
gdk/gdk_batop.c
gdk/gdk_private.h
gdk/gdk_string.c
monetdb5/modules/mal/tablet.c
Branch: Jul2021
Log Message:
A bunch of micro optimizations.
diffs (222 lines):
diff --git a/gdk/gdk_batop.c b/gdk/gdk_batop.c
--- a/gdk/gdk_batop.c
+++ b/gdk/gdk_batop.c
@@ -2093,6 +2093,40 @@ BATordered(BAT *b)
case TYPE_dbl:
BAT_ORDERED_FP(dbl);
break;
+ case TYPE_str:
+ for (BUN q = BUNlast(b), p = 1; p < q; p++) {
+ int c;
+ const char *p1 = BUNtail(bi, p - 1);
+ const char *p2 = BUNtail(bi, p);
+ if (p1 == p2)
+ c = 0;
+ else if (p1[0] == '\200') {
+ if (p2[0] == '\200')
+ c = 0;
+ else
+ c = -1;
+ } else if (p2[0] == '\200')
+ c = 1;
+ else
+ c = strcmp(p1, p2);
+ if (c > 0) {
+ b->tnosorted = p;
+ TRC_DEBUG(ALGO, "Fixed nosorted("
BUNFMT ") for " ALGOBATFMT " (" LLFMT " usec)\n", p, ALGOBATPAR(b), GDKusec() -
t0);
+ goto doreturn;
+ } else if (c < 0) {
+ assert(!b->trevsorted);
+ if (b->tnorevsorted == 0) {
+ b->tnorevsorted = p;
+ TRC_DEBUG(ALGO, "Fixed
norevsorted(" BUNFMT ") for " ALGOBATFMT "\n", p, ALGOBATPAR(b));
+ }
+ } else if (b->tnokey[1] == 0) {
+ assert(!b->tkey);
+ b->tnokey[0] = p - 1;
+ b->tnokey[1] = p;
+ TRC_DEBUG(ALGO, "Fixed nokey(" BUNFMT
"," BUNFMT") for " ALGOBATFMT "\n", p - 1, p, ALGOBATPAR(b));
+ }
+ }
+ break;
default: {
int (*cmpf)(const void *, const void *) =
ATOMcompare(b->ttype);
for (BUN q = BUNlast(b), p = 1; p < q; p++) {
diff --git a/gdk/gdk_private.h b/gdk/gdk_private.h
--- a/gdk/gdk_private.h
+++ b/gdk/gdk_private.h
@@ -365,7 +365,7 @@ ilog2(BUN x)
b && b->torderidx ? "O" : "", \
b ? b->timprints ? "I" : b->theap && b->theap->parentid &&
BBP_cache(b->theap->parentid) && BBP_cache(b->theap->parentid)->timprints ?
"(I)" : "" : ""
-#define BBP_BATMASK (128 * SIZEOF_SIZE_T - 1)
+#define BBP_BATMASK ((1 << (SIZEOF_SIZE_T + 5)) - 1)
#define BBP_THREADMASK 63
struct PROPrec {
diff --git a/gdk/gdk_string.c b/gdk/gdk_string.c
--- a/gdk/gdk_string.c
+++ b/gdk/gdk_string.c
@@ -115,7 +115,7 @@ strCleanHash(Heap *h, bool rebuild)
strhash = strHash(s);
off = strhash & GDK_STRHASHMASK;
newhash[off] = (stridx_t) (pos - extralen - sizeof(stridx_t));
- pos += strLen(s);
+ pos += strlen(s) + 1;
}
/* only set dirty flag if the hash table actually changed */
if (memcmp(newhash, h->base, sizeof(newhash)) != 0) {
@@ -136,7 +136,7 @@ strCleanHash(Heap *h, bool rebuild)
pos += pad + extralen;
s = h->base + pos;
assert(strLocate(h, s) != 0);
- pos += strLen(s);
+ pos += strlen(s) + 1;
}
}
#endif
@@ -170,7 +170,7 @@ strLocate(Heap *h, const char *v)
/* search the linked list */
for (ref = ((stridx_t *) h->base) + off; *ref; ref = next) {
next = (stridx_t *) (h->base + *ref);
- if (strCmp(v, (str) (next + 1) + extralen) == 0)
+ if (strcmp(v, (str) (next + 1) + extralen) == 0)
return (var_t) ((sizeof(stridx_t) + *ref + extralen));
/* found */
}
return 0;
@@ -256,7 +256,7 @@ strPut(BAT *b, var_t *dst, const void *V
const char *v = V;
Heap *h = b->tvheap;
size_t pad;
- size_t pos, len = strLen(v);
+ size_t pos, len = strlen(v) + 1;
const size_t extralen = h->hashash ? EXTRALEN : 0;
stridx_t *bucket;
BUN off, strhash;
@@ -293,7 +293,7 @@ strPut(BAT *b, var_t *dst, const void *V
do {
pos = *ref + sizeof(stridx_t) + extralen;
- if (strCmp(v, h->base + pos) == 0) {
+ if (strcmp(v, h->base + pos) == 0) {
/* found */
return *dst = (var_t) pos;
}
@@ -304,7 +304,7 @@ strPut(BAT *b, var_t *dst, const void *V
* linked list, so only look at single
* entry */
pos = *bucket + extralen;
- if (strCmp(v, h->base + pos) == 0) {
+ if (strcmp(v, h->base + pos) == 0) {
/* already in heap: reuse */
return *dst = (var_t) pos;
}
diff --git a/monetdb5/modules/mal/tablet.c b/monetdb5/modules/mal/tablet.c
--- a/monetdb5/modules/mal/tablet.c
+++ b/monetdb5/modules/mal/tablet.c
@@ -1223,6 +1223,16 @@ mkdfa(const unsigned char *sep, size_t s
return dfa;
}
+#ifdef __GNUC__
+/* __builtin_expect returns its first argument; it is expected to be
+ * equal to the second argument */
+#define unlikely(expr) __builtin_expect((expr) != 0, 0)
+#define likely(expr) __builtin_expect((expr) != 0, 1)
+#else
+#define unlikely(expr) (expr)
+#define likely(expr) (expr)
+#endif
+
static void
SQLproducer(void *p)
{
@@ -1267,7 +1277,7 @@ SQLproducer(void *p)
// we may be reading from standard input and may be out of input
// warn the consumers
if (ateof[cur] && partial) {
- if (partial) {
+ if (unlikely(partial)) {
tablet_error(task, rowno, int_nil, "incomplete
record at end of file", s);
task->b->pos += partial;
}
@@ -1275,7 +1285,7 @@ SQLproducer(void *p)
}
if (task->errbuf && task->errbuf[0]) {
- if (GDKerrbuf && GDKerrbuf[0]) {
+ if (unlikely(GDKerrbuf && GDKerrbuf[0])) {
tablet_error(task, rowno, int_nil, GDKerrbuf,
"SQLload_file");
/* TRC_DEBUG(MAL_SERVER, "Bailout on SQLload\n");*/
ateof[cur] = true;
@@ -1291,7 +1301,7 @@ SQLproducer(void *p)
s = task->input[cur];
base = end;
/* avoid too long records */
- if (end - s + task->b->len - task->b->pos >=
task->rowlimit[cur]) {
+ if (unlikely(end - s + task->b->len - task->b->pos >=
task->rowlimit[cur])) {
/* the input buffer should be extended, but 'base' is
not shared
between the threads, which we can not now update.
Mimick an ateof instead; */
@@ -1342,26 +1352,28 @@ SQLproducer(void *p)
} else {
/* check for correctly encoded UTF-8 */
if (nutf > 0) {
- if ((*e & 0xC0) != 0x80)
+ if (unlikely((*e & 0xC0) !=
0x80))
goto badutf8;
- if (m != 0 && (*e & m) == 0)
+ if (unlikely(m != 0 && (*e & m)
== 0))
goto badutf8;
m = 0;
nutf--;
- } else if ((*e & 0xE0) == 0xC0) {
- nutf = 1;
- if ((e[0] & 0x1E) == 0)
+ } else if ((*e & 0x80) != 0) {
+ if ((*e & 0xE0) == 0xC0) {
+ nutf = 1;
+ if (unlikely((e[0] &
0x1E) == 0))
+ goto badutf8;
+ } else if ((*e & 0xF0) == 0xE0)
{
+ nutf = 2;
+ if ((e[0] & 0x0F) == 0)
+ m = 0x20;
+ } else if (likely((*e & 0xF8)
== 0xF0)) {
+ nutf = 3;
+ if ((e[0] & 0x07) == 0)
+ m = 0x30;
+ } else {
goto badutf8;
- } else if ((*e & 0xF0) == 0xE0) {
- nutf = 2;
- if ((e[0] & 0x0F) == 0)
- m = 0x20;
- } else if ((*e & 0xF8) == 0xF0) {
- nutf = 3;
- if ((e[0] & 0x07) == 0)
- m = 0x30;
- } else if ((*e & 0x80) != 0) {
- goto badutf8;
+ }
}
/* check for quoting and the row
separator */
if (bs) {
@@ -1385,7 +1397,7 @@ SQLproducer(void *p)
if (*e == 0) {
partial = e - s;
/* found an incomplete record, saved for next
round */
- if (s+partial < end) {
+ if (unlikely(s+partial < end)) {
/* found a EOS in the input */
tablet_error(task, rowno, int_nil,
"record too long (EOS found)", "");
ateof[cur] = true;
@@ -1477,7 +1489,7 @@ SQLproducer(void *p)
/* move the non-parsed correct row data to the head of the next
buffer */
end = s = task->input[cur];
}
- if (cnt < task->maxrow && task->maxrow != BUN_NONE) {
+ if (unlikely(cnt < task->maxrow && task->maxrow != BUN_NONE)) {
char msg[256];
snprintf(msg, sizeof(msg), "incomplete record at end of
file:%s\n", s);
task->as->error = GDKstrdup(msg);
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
