Bugs item #2757110, was opened at 2009-04-13 02:04
Message generated for change (Comment added) made by stmane
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=482468&aid=2757110&group_id=56967

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: PF/compiler
Group: Pathfinder CVS Head
Status: Open
Resolution: None
Priority: 3
Private: No
Submitted By: Peter Boncz (boncz)
Assigned to: Jan Rittinger (tsheyar)
Summary: plan_subexpression slow

Initial Comment:
many MTest benchmarks/XBench queries, among which for instance 
benchmarks/XBench/DC/SD/q10.xq
fail due to timeout.

It appears many queries now take ages to compile, with the routine 
'plan_subexpression' being called in a very deep stack.

Is there a bug in the optimization algorithm, or has physical optimization by 
now become so ambitious we need a mechanism in the pf compiler to keep the 
search sapce under control (or stop searching after a time or effort limit). 

----------------------------------------------------------------------

>Comment By: Stefan Manegold (stmane)
Date: 2009-08-11 11:52

Message:
According to our TestWeb, no timeouts occur with any of the
pathfinder/benchmark/XBench tests (anymore):

http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/index.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTestsg103/index.html


----------------------------------------------------------------------

Comment By: Martin Kersten (mlkersten)
Date: 2009-08-06 16:36

Message:
It seems this has to wait for future improved versions.

----------------------------------------------------------------------

Comment By: Jan Rittinger (tsheyar)
Date: 2009-04-14 11:45

Message:
The physical plan generator was never designed to be especially
intelligent. It (stupidly) enumerates all possible plans. In the past I
tried to add some short-cuts to avoid too bad compilation performance.

With the recent optimization a lot of the ordering constraints, that
forced the plan space to stay small, were moved out of the way. Furthermore
the optimization option Y made the schemas of the operators much wider.
Mainly the latter change causes routine better_or_equal() to run into
serious performance problems as e.g., for query
benchmarks/XBench/DC/SD/q10.xq two plans are compared with 2061256 and
1374170 different orderings!!

I would guess that the problem becomes less dramatic again with the
MonetDB specific optimizations in place that restrict the schema width (see
also bug #2757104).

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=482468&aid=2757110&group_id=56967

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Monetdb-bugs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/monetdb-bugs

Reply via email to