I can't say I understand precisely what you tested, but I can tell you
that performance should have little to no dependence on fact ID
numbers. Jess's memory areas are densely indexed, and there are no
circumstances under which a linear search of all of memory is done.
On Feb 3, 2010, at 4:04 AM, Гильман Катя wrote:
Hello,
I have a question about the Rete in Jess,
does the ordering of facts in working memory affects the system
performance?
What i'm trying to understand, if e.g. i have 100 instances of the
same concept loaded in working memory,
will it affect much the performance if the matching rule satisfies
to the conept with order 2 or concept with order 90,
and what if i have 1000 instead ? In this case more concpets should
be checked?
I feel that it is more related to alpha memory implementation of the
algorithm, however if it is hashed then it should not affect much,
or am i wrong?
I did some measurements where i assert some amount of concepts and
measuring the response time it takes to apply the set of rules on
them, for instance, firstly i set 50 concepts and measuring the time
of 20 of them (rule is applicable only to the specific one, so
search has to be done), so for each measured concpet i have response
time. Then i repeat the procedure for more and more concepts. In
CLIPS it is noticable that response time increases when i have more
concepts and order of asserted facts is higher.
But for Jess situation is lilttle bit different, very first
measurement time is very high, comparing with others within
measurement set, and then with higher ordering it gets smaller. And
I don't understand why? I made measurements after everything was
asserted to working memory, so it is not the time of Rete network
construction.
Hopefully you can understand from this writing what i mean
Thank you very much
Best regards,
Kate
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users
y...@address.com'
in the BODY of a message to majord...@sandia.gov, NOT to the list
(use your own address!) List problems? Notify owner-jess-us...@sandia.gov
.
--------------------------------------------------------------------
---------------------------------------------------------
Ernest Friedman-Hill
Informatics & Decision Sciences, Sandia National Laboratories
PO Box 969, MS 9012, Livermore, CA 94550
http://www.jessrules.com
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users y...@address.com'
in the BODY of a message to majord...@sandia.gov, NOT to the list
(use your own address!) List problems? Notify owner-jess-us...@sandia.gov.
--------------------------------------------------------------------