Changeset: 2a4bb24f4e8a for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=2a4bb24f4e8a
Modified Files:
monetdb5/optimizer/opt_slicing.mx
Branch: default
Log Message:
Introduce simple partitioning heuristic
The bounds of the partitions are derived form the min/max values,
and include the other ranges,e.g. (nil,min,max,nil)
For more partitions an uniform distribution is assumed.
diffs (195 lines):
diff --git a/monetdb5/optimizer/opt_slicing.mx
b/monetdb5/optimizer/opt_slicing.mx
--- a/monetdb5/optimizer/opt_slicing.mx
+++ b/monetdb5/optimizer/opt_slicing.mx
@@ -42,7 +42,7 @@
opt_export str OPTgauges(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr
pci);
-#define _DEBUG_OPT_SLICE_
+/* #define _DEBUG_OPT_SLICE_ */
#define OPTDEBUGslicing if ( optDebug & ((lng)1 <<DEBUG_OPT_SLICING) )
#endif
@c
@@ -84,19 +84,6 @@
ValRecord bounds[MAXSITES];
} Gauge;
-static void
-OPTsliceGauges(MalBlkPtr mb, Gauge gauge, int peercnt)
-{
- /* use sampling to determine initial gauge setting */
- int i;
-
- (void) mb;
- for ( i=0; i < peercnt; i++){
- gauge.bounds[i].vtype = gauge.type;
- gauge.bounds[i].val.ival = 0;
- }
-}
-
static int
OPTfindPeer(str uri)
{
@@ -261,7 +248,7 @@
}
/* (k1,...kn):= remote.exec(conn,slicing,qry,version....) */
- snprintf(nme,BUFSIZ,"%s_subq",getFunctionId( getInstrPtr(mb,0)));
+ snprintf(nme,BUFSIZ,"%s_sub",getFunctionId( getInstrPtr(mb,0)));
q= newFcnCall(smb,remoteRef,execRef);
q->retc= q->argc= 0;
for (j=0; j < sig->retc; j++){
@@ -419,6 +406,13 @@
} else
if ( getFunctionId(q) == putName("exportValue",11))
getArg(q,8)= getArg(pack[0],0);
+ else
+ if ( getFunctionId(q) == NULL){
+ /* simple assignment */
+ assert(nrpack == q->retc);
+ for( k = 0; k < nrpack; k++)
+ getArg(q, q->retc + k) = getArg(pack[k],0);
+ }
pushInstruction(cmb,q);
@@ -484,7 +478,7 @@
str msg= MAL_SUCCEED;
/* define the sub query as part of the user space */
- snprintf(nme,BUFSIZ,"%s_subq%d",getFunctionId( getInstrPtr(mb,0)), pc);
+ snprintf(nme,BUFSIZ,"%s_qry%d",getFunctionId( getInstrPtr(mb,0)), pc);
s = newFunction(userRef, putName(nme, strlen(nme)),FUNCTIONsymbol);
if ( s == NULL)
return 0;
@@ -492,7 +486,6 @@
s->def = copyMalBlk(mb);
nmb = s->def;
getFunctionId( getInstrPtr(nmb,0)) = putName(nme,strlen(nme));
- insertSymbol(cntxt->nspace,s);
limit = nmb->stop;
old = nmb->stmt;
@@ -549,7 +542,6 @@
/* temporary extend signature with bounds */
nmb->stmt[0] =
pushArgument(nmb,getInstrPtr(nmb,0), gauge.lgauge);
nmb->stmt[0] =
pushArgument(nmb,getInstrPtr(nmb,0), gauge.hgauge);
- OPTsliceGauges(nmb, gauge, nrpeers);
}
if (strcmp(gauge.column, getVarConstant(nmb,
getArg(p,4)).val.sval) == 0) {
@@ -603,21 +595,18 @@
freeMalBlk(nmb);
return 0;
}
+ insertSymbol(cntxt->nspace,s);
q = 0;
/* supportive routines */
smb = OPTslicingStub(cntxt, mb, pc, nmb, gauge);
- if ( smb)
- printFunction(cntxt->fdout, smb, 0, LIST_MAL_STMT);
msg = OPTdeadcode(cntxt, smb, 0, 0);
if ( msg)
goto wrapup;
cmb = OPTslicingCntrl(cntxt,mb,pc,nmb, gauge);
- if ( cmb)
- printFunction(cntxt->fdout, cmb, 0, LIST_MAL_STMT);
chkProgram(cntxt->nspace, cmb);
if ( cmb->errors)
goto wrapup;
@@ -640,6 +629,7 @@
q->token = ASSIGNsymbol;
/* beware this instruction refers to cmb variables */
wrapup:
+#ifdef _DEBUG_OPT_SLICE_
mnstr_printf(cntxt->fdout,"#Grabbed part\n");
if ( nmb)
printFunction(cntxt->fdout, nmb, 0, LIST_MAL_STMT);
@@ -649,6 +639,7 @@
mnstr_printf(cntxt->fdout,"#cntrl\n");
if ( cmb)
printFunction(cntxt->fdout, cmb, 0, LIST_MAL_STMT);
+#endif
if (vec) GDKfree(vec);
return q;
@@ -674,7 +665,10 @@
if ( nrpeers == 0 )
OPTdiscover(cntxt);
mal_unset_lock(mal_contextLock,"optimizer.slicing");
+ /* nrpeers = GDKnr_threads ? GDKnr_threads:1;*/
+ /* to ease debugging now */
nrpeers = 2;
+ slicingLocal = 1;
/* modify the block as we go */
omb= copyMalBlk(mb);
@@ -809,20 +803,64 @@
@include optimizerWrapper.mx
@c
#include "opt_statistics.h"
+#include "algebra.h"
@:wrapOptimizer(slicing,OPT_CHECK_ALL)@
+@-
+@= histo
+ { @1 *vv;
+ vv= (@1 *) getArgReference(stk, pci, 0 );
+ *vv = @1_nil;
+ vx = (ptr) &max.val.@2;
+ BATmax(b, vx);
+ vv= (ptr) getArgReference(stk, pci, pci->retc -1 );
+ *vv = @1_nil;
+ vm = (ptr) &min.val.@2;
+ BATmin(b, vm);
+ step.val.@2 = (max.val.@2 - min.val.@2) / pci->retc-2;
+ if ( step.val.@2 == 0)
+ step.val.@2 = max.val.@2;
+ for ( i = 1; i < pci->retc-1; i++) {
+ vx= (ptr) getArgReference(stk, pci, i );
+ ((ValPtr) vx ) ->val.@2 = min.val.@2 + i * step.val.@2;
+ }
+ }
+@c
/* determine the initial gauge values */
+/* the min/max bounds are null values */
/* It is the heart of the approach and requires experimentation */
str
OPTgauges(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr pci)
{
BAT *b;
int bid = *(int*) getArgReference(stk,pci, pci->argc-1);
+ ValRecord min,max, step;
+ int i;
+ ptr vx,vm;
+
(void) cntxt;
(void) mb;
+ if ( pci->retc < 2)
+ throw(SQL,"sql.gauges","Not enough gauges");
b = BATdescriptor(bid);
if (b == NULL)
throw(SQL,"sql.gauges","Can not access BAT");
+ switch (ATOMstorage(b->ttype)) {
+ case TYPE_bte: @:histo(bte,btval)@ break;
+ case TYPE_sht: @:histo(sht,shval)@ break;
+ case TYPE_int: @:histo(int,ival)@ break;
+ case TYPE_wrd: @:histo(wrd,wval)@ break;
+ case TYPE_lng: @:histo(lng,lval)@ break;
+ case TYPE_oid: @:histo(oid,oval)@ break;
+ case TYPE_flt: @:histo(flt,fval)@ break;
+ case TYPE_dbl: @:histo(dbl,dval)@ break;
+ default:
+ throw(SQL,"sql.gauges","Can not slice column by value or oid");
+ }
+
+ for( i = 0; i < pci->retc; i++){
+ }
+
BBPreleaseref(bid);
return MAL_SUCCEED;
}
_______________________________________________
Checkin-list mailing list
Checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
