Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Acquiring the Hiarcs 7.32 engine with Fritz 5.32 (or the reverse)

Author: Harald Faber

Date: 00:37:18 08/06/99

Go up one level in this thread



On August 06, 1999 at 01:56:36, Steve Schooler wrote:

>Although my brief research makes Fritz an attractive buy, I am very intrigued by
>the following excerpt from the Hiarcs 7.32 features documentation:
>
>  The new 32Bit version of Hiarcs does not clear its hash tables between
>  moves which makes it fantastically suited for the automated backward
>  game analysis in the Hiarcs7.32 user interface. In backward analysis, a
>  program with persistent hash tables knows about the further game
>  continuation and thus looks much deeper.  And supported by position
>  learning you can show the program the dangers of a variation which it
>  would normally stumble into in analysis and thus force it to look for
>  superior alternatives.
>
>  Hiarcs7.32 accesses the Nalimov Tablebases in the search tree. This
>  boosts playing strength in simple endgames dramatically. In contrast,
>  Fritz5.32 only evaluates endgame databases at the root.
>
>Questions:
>
>1.  To experienced Hiarcs 7.32 users:  is the above excerpt for real?  I've
>never heard of "backward game analysis" before.  How/when is this used?

You have to differ. Backward game analysis can be:
a) Automatically analyzing (at least) one game from the final position backwards
to the beginning. This feature is implemented in all Fritz-family-programs
b) like described above, you make some moves on the board while Hiarcs is in
analyze mode. Herein Hiarcs eval finds some loss or win, and when you take back
the moves one by one, you see Hiarcs eval adapted the "learned". Example: You
have a position where Hiarcs has +0.23. You make a move (e.g. Nd5), wait 10-30
(or even more) seconds and Hiarcs changes the eval up to +3.42. If you now take
back that last move you made, Hiarcs immediately will show the PV with the best
move (Nd5) with the eval +3.42.

BTW I am told that Shredder uses this method since Version 2.

>To simplify responses to my remaining questions, please ASSUME that Hiarcs 7.32
>search engine has an advantage re above excerpt, and that Fritz has faster
>nodes/sec search engine.
>
>2.  Rather than spending (approx.) $100 to acquire both, is it possible to
>economically get the advantages of both under one umbrella (i.e. Fritz 5.32 with
>the Hiarcs 7.32 search engine as an add-on, or the REVERSE)?  This question
>really has two parts:
>
>2a. Can the Hiarcs 7.32 search engine (only, with no interface) be purchased for
>significantly less than the $48+ of the full Hiarcs 7.32 package, and then
>"plugged into" Fritz 5.32 as an add-on.  Also included in question 2a is the
>REVERSE, re buying and plugging Fritz engine into Hiarcs.

There is only full Hiarcs7.32 incl. GUI, no engine itself. Of course you may add
this engine out of the full package to the other Fritz-family programs.

>2b. Will the plug-in strategy of Question 2a truly consolidate both advantages?
>For example, the above excerpt suggests that the Hiarcs 7.32 interface allows
>"automated backward game analysis", while the Fritz interface does not.  I infer
>that plugging the Fritz engine into the Hiarcs 7.32 package will achieve my
>goal, while the reverse will not.  IS THIS TRUE?

No. So far only Hiarcs handles the hashtables that way. It does not depend on
the GUI but on the engine so adding Fritz5.32 engine to Hiarcs7.32 wouldn't
provide this function.

>3.  In the rec.games.chess.computers newsgroup, someone informally contrasted
>the two search engines (Fritz 5.32 vs Hiarcs 7.32) by calling Fritz a tactical
>monster, and indicating that Hiarcs more often "agrees" with the move actually
>chosen by the Grandmaster.  Perhaps this relates to Hiarcs' "backward game
>analysis" feature.  Intriguing, REQUEST FEEDBACK ON THIS.

Hiarcs is also tactically strong. Maybe not as strong as Fritz but strong
enough.

>4.  Taking everything above with a large boulder of salt:  if it is all true,
>then I infer that Fritz is consistently superior in analyzing a "static"
>position (i.e. setup a position and then say analyze).  REQUEST FEEDBACK ON
>THIS.

Static positions? No, you have to give Fritz DYNAMIC positions.
For positional play get MChessPro, Rebel, Shredder or Hiarcs.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.