Re: large databases
- From: Jan Wielemaker <jan@xxxxxxxxxxxxxxxxxxx>
- Date: 21 May 2008 17:28:56 GMT
On 2008-05-21, Simon Strobl <Simon.Strobl@xxxxxxxxx> wrote:
The Prolog system should of course be able to handle it, but most
importantly, make sure most of your database predicates are properly
indexed.
O.k. thanks for the tip. At the time being, however, my problem is not
the time Prolog uses for making use of facts. My problem is that
Prolog does not manage to load the facts. As I said, Yap crashes
(although enough RAM was available). Swi did not crash, but it did not
succeed to load the file in several days. I have now started a new Swi-
process. I will wait a week or two in order to see if Swi can load the
file.
SWI should load this *much* faster, so something is wrong. What OS are
you using and is this a 32-bit or 64-bit version? It is possible SWI
crashed too; sometimes it recovers poorly on fatal errors. You can
avoid recovery using pl --nosignals. Just to show:
?- tell('x.pl'), forall(between(1, 10000000, X), format('~q.~n', [a(X)])), told.
ct (~) 4_> ls -lh x.pl
-rw-r--r-- 1 jan users 114M 2008-05-21 19:23 x.pl
?- time([x]).
% x compiled 56.44 sec, 1,040,000,720 bytes
% 160,000,369 inferences, 56.44 CPU in 57.37 seconds (98% CPU, 2834875 Lips)
This is using 64-bit Linux version on AMD Athlon 5400+. That gives some
idea about the loading time to expect. Of course, these are very simple
facts.
--- Jan
.
- Follow-Ups:
- Re: large databases
- From: Simon Strobl
- Re: large databases
- References:
- large databases
- From: Simon Strobl
- Re: large databases
- From: Jan Wielemaker
- Re: large databases
- From: Simon Strobl
- large databases
- Prev by Date: Visual Prolog Application and Language Conference 2008 in Saint Petersburg
- Next by Date: Re: Networking in Prolog - a survey
- Previous by thread: Re: large databases
- Next by thread: Re: large databases
- Index(es):
Relevant Pages
|
