Pupsi2 0.08 released.

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

jesper_nielsen

Pupsi2 0.08 released.

Post by jesper_nielsen »

Hi All,

The release you have all been waiting for! Well at least Dadi Johnson waited! :D

A hopefully IDeA compliant version of Pupsi2! Yay! :?

It should rightfully be named, "The Phoenix version", since it has risen from the ashes.

I worked long trying stuff, but nothing gave any progress. The best version was the ALPHA 7 version. After that the versions became worse and worse, in spite of my conviction that the changes had to be good. Somewhere around version ALPHA 14, I decided that anough was enough! Out with all the useless changes! Back to ALPHA 7!

At that point the server with my svn repository crashed and burned! :(

It was hosted offline at a friend of mine! :cry:

And also at this point I was forced to change dotnet from 1.1 to 3.5, because of upgrade to the compiler I use. I have been trying to avoid this, since all indications are that Pupsi loses 5% performance going away from 1.1. :(

After many many tries to find the older source code again, I managed to find ALPHA 8. Then I reverse engineered the executable for ALPHA 7, and applied the changes to get something that was kinda like ALPHA 7.

So this release is kind of a "lets start from here" type release.

From this version forward, I am using the brilliant "cutechess-cli" tool for running fast test games. I use many starting positions generated ala Hyatt. But still testing will be painfully slow, since I have very limited access to CPU power.

So to make a long post .... longer .... here it is: 0.08

Tip for people stuck looking for improvements. Try testing a change to a parameter. If it makes the engine worse, you at least get to be slightly optimistic about changing the parameter in the opposite direction! :D

Kind regards,
Jesper
User avatar
Graham Banks
Posts: 41455
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Pupsi2 0.08 released.

Post by Graham Banks »

Thanks Jesper. :D
gbanksnz at gmail.com
jesper_nielsen

Re: Pupsi2 0.08 released.

Post by jesper_nielsen »

Oops! Forgot to mention that it is available for download at Leo's site.

http://wbec-ridderkerk.nl/

Thank you Leo, for hosting it!

Kind regards,
Jesper
User avatar
beachknight
Posts: 3533
Joined: Tue Jan 09, 2007 8:33 pm
Location: Antalya, Turkey

Re: Pupsi2 0.08 released.

Post by beachknight »

Thanks. One of my favourites.

Beat 'em all, Pupsi.

Best,
hi, merhaba, hallo HT
User avatar
Marek Soszynski
Posts: 582
Joined: Wed May 10, 2006 7:28 pm
Location: Birmingham, England

Re: Pupsi2 0.08 released.

Post by Marek Soszynski »

jesper_nielsen wrote:A hopefully IDeA compliant version of Pupsi2! Yay!
Yay! indeed.
Marek Soszynski
swami
Posts: 6640
Joined: Thu Mar 09, 2006 4:21 am

Re: Pupsi2 0.08 released.

Post by swami »

Pupsi 0.08
by Jesper Nielsen, Denmark.

Strategic Test Suite Conditions:
Arena GUI
10 seconds per position
700 positions
Core2Quad 32 bits, Q6600, 2 GB RAM, 2.4GHZ
Engine uses 113 Mb Hash.
Subject-wise Scores:
STS (v1.0) - Undermining:
71/100, Grade: A-

STS (v2.0) - Open Files and Diagonals:
72/100, Grade: A-

STS (v3.0) - Knight Outposts/Centralization/Repositioning:
57/100, Grade: C+

STS (v4.0) - Square Vacancy:
64/100, Grade: B

STS (v5.0) - Bishop vs Knight:
60/100, Grade: B

STS (v6.0) - Re-Capturing:
60/100, Grade: B+

STS (v7.0) - Offer of Simplification:
62/100, Grade: B
Overall Performance:
Total Score: 446/700

Overall Average: 63.71 %

Grade: B
mhalstern
Posts: 484
Joined: Wed Nov 18, 2009 1:09 am

Re: Pupsi2 0.08 released.

Post by mhalstern »

Congratulations.

I am using it with my Fritz 11 Interface. I'll report the elostat results in a few weeks.

I noticed one thing: When the engine first starts to analyze, something, the reported depth count is unusually low for a while. Then, when it move, the numbers jump up much higher.
User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 8:30 pm
Location: Chicago, Illinois, USA

Re: Pupsi2 0.08 released.

Post by michiguel »

Marek Soszynski wrote:
jesper_nielsen wrote:A hopefully IDeA compliant version of Pupsi2! Yay!
Yay! indeed.
What is that? IDeA compliant?

Miguel
User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

Re: Pupsi2 0.08 released.

Post by Matthias Gemuh »

michiguel wrote:
Marek Soszynski wrote:
jesper_nielsen wrote:A hopefully IDeA compliant version of Pupsi2! Yay!
Yay! indeed.
What is that? IDeA compliant?

Miguel
Interactive Deep Analysis in Rybka GUI.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
jesper_nielsen

Re: Pupsi2 0.08 released.

Post by jesper_nielsen »

mhalstern wrote:Congratulations.

I am using it with my Fritz 11 Interface. I'll report the elostat results in a few weeks.

I noticed one thing: When the engine first starts to analyze, something, the reported depth count is unusually low for a while. Then, when it move, the numbers jump up much higher.
Hmmm.... that sounds wierd.... I haven't noticed it myself.

I report depth in two situations:
1. Whenever the PV changes.
2. Whenever an iteration completes.

So you might see something like "depth 9 seldepth 14" along with a pv.
And then a bit later "depth 9 seldepth 16" sometimes later when the iteration completes, if searching alternative moves takes it to ply 16 using extentions.

Kind regards,
Jesper