BB's Rybka/Ippolit comparison

General discussion about computer chess...
BB+
Posts: 1484
Joined: Thu Jun 10, 2010 4:26 am

Re: BB's Rybka/Ippolit comparison

Post by BB+ » Tue Jun 15, 2010 5:34 am

I wrote: "I think the castling code given is one of the more easy things to understand."

Oops, I forgot that what I gave in the PDF had the %rip stuff and omitted the actual location to which the relative instruction pointer points (this was done for formatting purposes). The gdb dump looks more like this (though here I leave out my comments):

Code: Select all

0x00482a00:     xor    %eax,%eax
0x00482a02:     cmp    $0x6,%ecx
0x00482a05:     mov    %eax,2392757(%rip)        # 0x6cacc0
0x00482a0b:     jne    0x482a84
0x00482a0d:     xorq   $0xa0,2392584(%rip)        # 0x6cac20
0x00482a18:     xorq   $0xa0,2392637(%rip)        # 0x6cac60
0x00482a23:     xorq   $0xa0,2392698(%rip)        # 0x6caca8
0x00482a2e:     xorq   $0x9,2392682(%rip)        # 0x6caca0
0x00482a36:     addl   $0x5a0000,2392712(%rip)        # 0x6cacc8
0x00482a40:     mov    %eax,2392310(%rip)        # 0x6cab3c
0x00482a46:     mov    $0x8000800000000000,%rax
0x00482a50:     movl   $0x8,2392282(%rip)        # 0x6cab34
0x00482a5a:     xor    %rax,2392655(%rip)        # 0x6cacb0
0x00482a61:     mov    $0x800100000,%rax
0x00482a6b:     xor    %rax,2392614(%rip)        # 0x6cac98
0x00482a72:     mov    $0xd8b3287ea544969,%rax
0x00482a7c:     xor    %rax,2392653(%rip)        # 0x6cacd0
0x00482a83:     retq
0x00482a84:     xorq   $0x9,2392468(%rip)        # 0x6cac20
0x00482a8c:     xorq   $0x9,2392524(%rip)        # 0x6cac60
0x00482a94:     xorq   $0x9,2392588(%rip)        # 0x6caca8
0x00482a9c:     xorq   $0x201,2392561(%rip)        # 0x6cac98
0x00482aa7:     xorq   $0x10000400,2392558(%rip)        # 0x6caca0
0x00482ab2:     addl   $0x820000,2392588(%rip)        # 0x6cacc8
0x00482abc:     mov    %eax,2392158(%rip)        # 0x6cab20
0x00482ac2:     mov    $0x80000080,%eax
0x00482ac7:     movl   $0x8,2392155(%rip)        # 0x6cab2c
0x00482ad1:     xor    %rax,2392536(%rip)        # 0x6cacb0
0x00482ad8:     mov    $0xaaaff37267ceded3,%rax
0x00482ae2:     xor    %rax,2392551(%rip)        # 0x6cacd0
0x00482ae9:     retq
One gets sufficiently used to the repeated appearances of 0x6ca... locations, to be able to recognise them as bitboards, or the board, or static values, or the hash, etc.

BB+
Posts: 1484
Joined: Thu Jun 10, 2010 4:26 am

Re: BB's Rybka/Ippolit comparison

Post by BB+ » Tue Jun 15, 2010 5:40 am

Marco could not believe how Ippo can get away with such an aggressive (logarithmic) LMR
I'm not sure I would say that the "aggressive" part of LMR is the "logarithmic" scaling (as few moves really show up in the tail). Just a step function (at ALL nodes) from 2 to 3, then 3 to 4 half-ply at various places would seem to give the same effect. The location of the first step seems important, and the second not so important, though not irrelevant. At least this is what I was able to discern (from some of my own testing, and conversation with others -- and it can all be engine independent). Certainly others have been surprised as to how "conservative" the LMR in R3 really is, not having the first up-step until the 12th move.

User avatar
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: BB's Rybka/Ippolit comparison

Post by kingliveson » Tue Jun 15, 2010 5:53 am

BB+ wrote: .
.
.
I think I said, even when posting on the Rybka forum back in January or whenever, that I found it impossible to conceive that the IPPOLIT maker(s) did not know how Rybka worked; but given that the claim was that they were "clones", and that "code" was copied, I chose to emphasize the counter-evidence to this. I would also hope that a careful reading of the report would make it clear that IPPOLIT must have "come from" Rybka in some sense, though not at the level of "code" [as an aside, I tend to agree with Alan Sassler and others that reverse engineering is a legitimate form of discovery].
.
.
.
First, hope this server is regularly backed up. We know this thread would not have been allowed on Capitalist Controlled Chess server. To your last point based on the analysis and report -- in plain English, whoever authored Ippolit has a clear and working knowledge of Rybka (probably version 3) through dis-assembled code. I am quite convinced. The process by which it was done, by all account is legal. Ideas are not copyrightable, but rather expression of ideas.

IM Vasik Rajlich knows what occurred. He alluded to it in vague published (with permission) email statement just the other day. There is no legal case to be made, and he does not have what would stand in court. This should explain absence of evidence after a year long debate over "clone" accusation -- unlike the case against Strelka. In retrospect, Vas probably did wisely keeping silent. I am sure he was ticked off knowing the competition is about to stiffen. This situation is not unique and occurs in our technology world more than some would prefer.

Where do we go from here? One lesson to be learned at least is try to see things from more than a single point of view. Division that occurred in the chess community over this debate was/is unnecessary. The problem of course were fanboys and devout worshipers with no technical background to comprehend what has happened. Others were biased based on personal interest. I do not expect things to change much though since a segment of the community is cemented to its conclusion. Just remember that taking ideas from other programs did not start with Ippolit.
05.12.2005, Vasik Rajlich wrote: Yes, the publication of Fruit 2.1 was huge. Look at how many engines took a massive jump in its wake: Rybka, Hiarcs, Fritz, Zappa, Spike, List, and so on. I went through the Fruit 2.1 source code forwards and backwards and took many things.
PAWN : Knight >> Bishop >> Rook >>Queen

govert
Posts: 9
Joined: Thu Jun 10, 2010 10:47 am
Real Name: Martin Helmer

Re: BB's Rybka/Ippolit comparison

Post by govert » Tue Jun 15, 2010 7:27 am

LiquidNitrogen wrote:There's something else that may have escaped notice.

The writer of this document must have been a published author of an International Computer Games Association Journal article. I've seen this "format" before, and as a 3-time author myself, I recognized it right away.

Compare the layout of the PDF file to these ICGA Journal articles I authored:

http://www.GothicChess.com/80.pdf

http://www.GothicChess.com/7_piece.pdf

I'm just saying, this author was very knowledgeable and an experienced technical writer as well.
I'll bet a 100 dollars that the answer to that riddle is LaTex, which is used by any geek with self respect when writing a white paper.
BB, do I win?

User avatar
Kevin Frayer
Posts: 53
Joined: Fri Jun 11, 2010 8:07 pm
Real Name: Kevin Frayer
Location: Vincennes IN USA
Contact:

Re: BB's Rybka/Ippolit comparison

Post by Kevin Frayer » Tue Jun 15, 2010 8:21 am

BB I would like to thank you for this amazing work.

It is a somewhat sad thing, that perhaps the most important contribution to computer chess this year will be your proof that Vasik Rajlick and his followers have indeed been wrong in casting aspersions on the open source engine Ippolit.

Hundreds of computer chess enthusiast have been either willingly or unwillingly caught up in this controversy and it has given the general public a negative impression of our hobby. I wish it was over, but of course it is not. So long as commercial interests still see it profitable to use censorship and black listing tactics, the fight must go on.

The ultimate goal is the advancement of chess playing software. We all would be willing to pay for programers and developers to give us what we want, small but steady improvements, bug fixes and honesty.

Again, thank you for the White Paper.
Sic gorgiamus allos subjectatos nunc.

oudheusa
Posts: 20
Joined: Tue Jun 15, 2010 9:08 am

Re: BB's Rybka/Ippolit comparison

Post by oudheusa » Tue Jun 15, 2010 10:02 am

Rebel wrote:Finished reading the BB document. A couple of comments:
Hi Ed, great to see you here. The virus never dies, does it? ;)

User avatar
thorstenczub
Posts: 592
Joined: Wed Jun 09, 2010 12:51 pm
Real Name: Thorsten Czub
Location: United States of Europe, germany, NRW, Lünen
Contact:

Re: BB's Rybka/Ippolit comparison

Post by thorstenczub » Tue Jun 15, 2010 10:54 am

kingliveson wrote: Where do we go from here? One lesson to be learned at least is try to see things from more than a single point of view. Division that occurred in the chess community over this debate was/is unnecessary. The problem of course were fanboys and devout worshipers with no technical background to comprehend what has happened. Others were biased based on personal interest. I do not expect things to change much though since a segment of the community is cemented to its conclusion. Just remember that taking ideas from other programs did not start with Ippolit.
Why do rating lists exclude the "clones" ?
is it really UP TO THEM to decide about moral issues or court cases ?
Is this "protection" really helping computerchess.

all those "Fanboys", waterbearers, involved people, they all wash dirty laundry.
from vas rajlich we have seen NO evidence. Nothing. just a few statements, very
unclear. all the other stuff was done by the 2nd line of water-bearers. people
who debate about this or that, stuff that is not up to them.
but they like to do so, as if they would have to decide.

why did the elo guys exluded the so called clone programs ?
because this whole scene of bean counters needs something to justify their own
existence.

bob hyatt was attacked, christophe theron, steve b, and and and.
so many people have been hunted by the computerchess-stasi.
attacked for WHICH reason ?
because those guys "supported" the clones ?!
there was never any evidence that these programs are clones.
but the attacks against people did happen.
the whole community splitted itself in fangroups believing in the holy words
of their god rajlich, and in people who did not believe in the strange
statements by rajlich, who were completely contradictional.

sorry but these censorship and witch hunt showed what we can expect from some
people.
blind support without any evidence.
with these behaviour you can support any corrupt regime, you can
support any crime or holocaust if you have people who blindly follow the master
with their own brain switched off.

User avatar
Rebel
Posts: 515
Joined: Wed Jun 09, 2010 7:45 pm
Real Name: Ed Schroder

Re: BB's Rybka/Ippolit comparison

Post by Rebel » Tue Jun 15, 2010 11:00 am

oudheusa wrote:Hi Ed, great to see you here. The virus never dies, does it? ;)
Every time my shrink declares me cured the virus mutates itself :lol:

Ed

Jeremy Bernstein
Site Admin
Posts: 1226
Joined: Wed Jun 09, 2010 7:49 am
Real Name: Jeremy Bernstein
Location: Berlin, Germany
Contact:

Re: BB's Rybka/Ippolit comparison

Post by Jeremy Bernstein » Tue Jun 15, 2010 12:19 pm

kingliveson wrote:Where do we go from here?
Ideally, we get some independent confirmation from a reliable source that the data presented by BB is legit. I don't mean to be in any way disrespectful when I say that, BB. But given the events of the last months, it seems wise to verify claims, even (especially) those which resonate with one's own gut feeling. I wonder if Zach could be convinced to at least spot-check a few things.

Jeremy

govert
Posts: 9
Joined: Thu Jun 10, 2010 10:47 am
Real Name: Martin Helmer

Re: BB's Rybka/Ippolit comparison

Post by govert » Tue Jun 15, 2010 1:44 pm

Jeremy Bernstein wrote:
kingliveson wrote:Where do we go from here?
Ideally, we get some independent confirmation from a reliable source that the data presented by BB is legit. I don't mean to be in any way disrespectful when I say that, BB. But given the events of the last months, it seems wise to verify claims, even (especially) those which resonate with one's own gut feeling. I wonder if Zach could be convinced to at least spot-check a few things.

Jeremy
Also, I think we should get to the bottom of this anonymity thingy;

Assuming IPPO & co. are legal, why does the author(s) stay anonymous, and could something be done to get them to reveal their identities?

Post Reply