Changeset: 6cd501a905c3 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=6cd501a905c3
Modified Files:
monetdb5/modules/mal/joinpath.c
Branch: default
Log Message:
Revert chain join code
The initial thought that avoiding the intermediates in simple
oid-chain joins does not materialize in sf10, sf100. It is
even somewhat slower, most likely due to the interference
of pages from the first argument being swapped out when
you traverse the dependencies.
diffs (134 lines):
diff --git a/monetdb5/modules/mal/joinpath.c b/monetdb5/modules/mal/joinpath.c
--- a/monetdb5/modules/mal/joinpath.c
+++ b/monetdb5/modules/mal/joinpath.c
@@ -150,96 +150,6 @@ ALGjoinCost(Client cntxt, BAT *l, BAT *r
return cost;
}
-/*
- * The joinChain assumes a list of OID columns ending in a projection column.
- * It is built from leftfetchjoin operations, which allows for easy chaining.
- * No intermediates are needed and not multistep cost-based evaluation
- */
-
-#define MAXCHAINDEPTH 256
-
-static BAT *
-ALGjoinChain(Client cntxt, int top, BAT **joins)
-{
- BAT *bn = NULL;
- oid lo, hi, oc;
- BATiter iter[MAXCHAINDEPTH];
- int i, pcol= top -1, td = 1, ts = 1;
- BUN cnt=0, cap=0, empty=0, size[MAXCHAINDEPTH];
- ssize_t offset[MAXCHAINDEPTH];
- const void *v;
-
- //#undef ALGODEBUG
- //#define ALGODEBUG if(1)
- (void) cntxt;
-
- for ( i =0; i< top ; i++){
- if( (cnt = BATcount(joins[i]) ) > cap)
- cap = BATcount(joins[i]);
- empty += cnt == 0;
- iter[i] = bat_iterator(joins[i]);
- size[i] = BATcount(joins[i]);
- offset[i] = BUNfirst(joins[i])-joins[i]->hseqbase;
- ALGODEBUG {
- mnstr_printf(cntxt->fdout,"#%d types [%d, %d] "BUNFMT"
"SSZFMT" \n",i, joins[i]->htype, joins[i]->ttype, size[i], offset[i]);
- }
- if (i<(top-1))
- td &= joins[i]->tdense;
- if (i<(top-1))
- ts &= joins[i]->tsorted;
- }
-
- bn = BATnew( TYPE_void, joins[pcol]->ttype?joins[pcol]->ttype:TYPE_oid,
cap, TRANSIENT);
- if( bn == NULL){
- GDKerror("joinChain" MAL_MALLOC_FAIL);
- return NULL;
- }
- /* be optimistic, inherit the properties */
- BATsettrivprop(bn);
- BATseqbase(bn, 0);
- if (empty)
- return bn;
-
- bn->tkey = td&&joins[pcol]->tkey;
- bn->tdense = 0;
- bn->tsorted = ts&&joins[pcol]->tsorted;
- bn->trevsorted = 0;
- bn->T->nil = 0;
- bn->T->nonil = joins[pcol]->T->nonil;
-
- cnt = 0;
- for (lo = 0, hi = lo + BATcount(joins[0]); lo < hi; lo++) {
- oc = *(oid *) BUNtail(iter[0], lo);
- for(i = 1; i < pcol; i++)
- if (oc + offset[i] < size[i]) {
- v = BUNtail(iter[i], (BUN) (oc + offset[i]));
- oc = *(oid*) v;
- if (oc == oid_nil)
- goto bunins_failed;
- }
-
- if (i != pcol)
- continue;
- // update the join result and keep track of properties
- if (oc + offset[pcol] < size[pcol]){
- v = BUNtail(iter[pcol], (BUN) (oc + offset[pcol]));
- bunfastapp(bn,v);
- cnt++;
- }
- bunins_failed:
- ;
- }
- BATsetcount(bn, cnt);
- bn->hrevsorted = (BATcount(bn) <=1);
-
- // release the chain
- for ( i =0; i< top ; i++)
- BBPunfix(joins[i]->batCacheid);
-
- if (bn && !(bn->batDirty&2)) BATsetaccess(bn, BAT_READ);
-
- return bn;
-}
static BAT *
ALGjoinPathBody(Client cntxt, int top, BAT **joins, int flag)
@@ -362,7 +272,7 @@ ALGjoinPathBody(Client cntxt, int top, B
str
ALGjoinPath(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr pci)
{
- int i,top=0, chain = 1, empty = 0;
+ int i,top=0, empty = 0;
bat *bid;
bat *r = getArgReference_bat(stk, pci, 0);
BAT *b, **joins = (BAT**)GDKmalloc(pci->argc*sizeof(BAT*));
@@ -393,14 +303,11 @@ ALGjoinPath(Client cntxt, MalBlkPtr mb,
empty += BATcount(b) == 0;
joins[top++] = b;
}
- /* detect easy left-right oid chain joins */
- chain = BATcount(joins[0]) < BATcount(joins[top-1]) && top <
MAXCHAINDEPTH;
- chain = 0; // disabled for the moment, because it is not robust yet
ALGODEBUG{
char *ps;
ps = instruction2str(mb, 0, pci, LIST_MAL_ALL);
- fprintf(stderr,"#joinpath [%s] %s\n", (ps ? ps : ""),
chain?"chain":"diverse");
+ fprintf(stderr,"#joinpath %s\n", (ps ? ps : ""));
GDKfree(ps);
}
if ( empty){
@@ -415,8 +322,6 @@ ALGjoinPath(Client cntxt, MalBlkPtr mb,
BATsettrivprop(b);
} else if (getFunctionId(pci) == leftjoinPathRef) {
b = ALGjoinPathBody(cntxt,top,joins, 0);
- } else if (chain && top < MAXCHAINDEPTH) {
- b = ALGjoinChain(cntxt,top,joins);
} else {
b = ALGjoinPathBody(cntxt,top,joins, 3);
}
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
