Moderators: hgm, Rebel, chrisw
No paranoia. Simply a desire to test the engine of anybody whoDe Vos W wrote:
And I
do not want to test an engine that I have doubts about the author's
honesty concerning his/her engine. I have done so a few times already
and wish to avoid that.
Is this paranoia on a finer scale?
I believe Robert's response to the disassembled code above was this:K I Hyams wrote:I hold exactly the same view that you hold and I am in much the same position as you are.
If I were to release an engine that was likely to arouse the sort of controversy that currently prevails, I would be very careful to be absolutely open and extremely accurate about the origins of that engine. If I took a block of code, I would acknowledge it. The underlined and italicised quote is from the Houdart site.
Without many ideas from the excellent open source chess engines Ippolit/Robbolito, Stockfish and Crafty, Houdini would not nearly be as strong as it is now.
Rajlich acknowledged, somewhat grudgingly, an influence from Fruit. Robert acknowledges, less grudgingly, influences from Ippolit/Robbolito, Stockfish and Crafty. However, it may be that he has been “economical with the truth”. The quote below is from Lance Perkins, author of Thinker. One might wonder whether the claim that “This Houdini is just a cheap compile of the Ippo/Robo source code” is an example of Lance also being “economical with the truth”, however if the material that he published is genuine, it would seem to indicate to this layman that Robert has taken more than “ideas” from Robolito.CThinker wrote:This Houdini is just a cheap compile of the Ippo/Robo source code.
Below is one example of a decompiled Robolito and Houdini. The entire decompiled code is like this. You will find a one-to-one equivalence between the two.
I even found a few bugs in Robo/Ippo that were also in Houdini. Really funny. I will not disclose the bug. This way, I have an easy method of spotting these types of cheap tricks to compile Robo/Ippo and release it as something new.
Whoever did this Houdini thing is clueless about Robo/Ippo.
From the function...Decompiled Houdini:Code: Select all
static void Material_calculation( int c ) { int white_queen, black_queen, white_rook, black_rook, white_clear, black_clear, white_dark, black_dark, white_knight, black_knight, white_pawn, black_pawn, n, value, white_bishop, black_bishop; int ca, white_load, black_load, phase, va1, va2, va3, va4; uint64 va; n = c; white_queen = n % 2; n /= 2; black_queen = n % 2; n /= 2; white_rook = n % 3; n /= 3; black_rook = n % 3; n /= 3; white_clear = n % 2; n /= 2; white_dark = n % 2; n /= 2; black_clear = n % 2; n /= 2; black_dark = n % 2; n /= 2; white_knight = n % 3; n /= 3; black_knight = n % 3; n /= 3; white_pawn = n % 9; n /= 9; black_pawn = n % 9; white_bishop = white_clear + white_dark; black_bishop = black_clear + black_dark; va = value_value(white_pawn, white_knight, white_bishop, white_clear, white_dark, white_rook, white_queen, black_pawn, black_knight, black_bishop, black_clear, black_dark, black_rook, black_queen); phase = PHASE_MINOR * (white_knight + white_bishop + black_knight + black_bishop) + PHASE_ROOK * (white_rook + black_rook) + PHASE_QUEEN * (white_queen + black_queen); // etc., etc. }Decompiled Robo:Code: Select all
:00428364 8BE8 mov ebp, eax :00428366 81E501000080 and ebp, 80000001 :0042836C 7905 jns 00428373 :0042836E 4D dec ebp :0042836F 83CDFE or ebp, -00000002 :00428372 45 inc ebp --------- :00428373 99 cdq :00428374 2BC2 sub eax, edx :00428376 D1F8 sar eax, 1 :00428378 8BD8 mov ebx, eax :0042837A 81E301000080 and ebx, 80000001 :00428380 7905 jns 00428387 :00428382 4B dec ebx :00428383 83CBFE or ebx, -00000002 :00428386 43 inc ebx --------- :00428387 99 cdq :00428388 2BC2 sub eax, edx :0042838A 8BC8 mov ecx, eax :0042838C D1F9 sar ecx, 1 :0042838E B856555555 mov eax, 55555556 :00428393 F7E9 imul ecx :00428395 8BC2 mov eax, edx :00428397 C1E81F shr eax, 1F :0042839A 03C2 add eax, edx :0042839C 8D1440 lea edx, dword[eax+2*eax] :0042839F 2BCA sub ecx, edx :004283A1 894C2418 mov dword[esp+18], ecx :004283A5 8BC8 mov ecx, eax :004283A7 B856555555 mov eax, 55555556 :004283AC F7E9 imul ecx :004283AE 8BC2 mov eax, edx :004283B0 C1E81F shr eax, 1F :004283B3 03C2 add eax, edx :004283B5 8D1440 lea edx, dword[eax+2*eax] :004283B8 2BCA sub ecx, edx :004283BA 8BF8 mov edi, eax :004283BC 81E701000080 and edi, 80000001 :004283C2 894C2410 mov dword[esp+10], ecx :004283C6 7905 jns 004283CD :004283C8 4F dec edi :004283C9 83CFFE or edi, -00000002 :004283CC 47 inc edi --------- :004283CD 99 cdq :004283CE 2BC2 sub eax, edx :004283D0 D1F8 sar eax, 1 :004283D2 8BC8 mov ecx, eax :004283D4 81E101000080 and ecx, 80000001 :004283DA 897C2448 mov dword[esp+48], edi :004283DE 7905 jns 004283E5 :004283E0 49 dec ecx :004283E1 83C9FE or ecx, -00000002 :004283E4 41 inc ecx --------- :004283E5 99 cdq :004283E6 2BC2 sub eax, edx :004283E8 D1F8 sar eax, 1 :004283EA 894C2434 mov dword[esp+34], ecx :004283EE 8BC8 mov ecx, eax :004283F0 81E101000080 and ecx, 80000001 :004283F6 7905 jns 004283FD :004283F8 49 dec ecx :004283F9 83C9FE or ecx, -00000002 :004283FC 41 inc ecx --------- :004283FD 99 cdq :004283FE 2BC2 sub eax, edx :00428400 D1F8 sar eax, 1 :00428402 894C2428 mov dword[esp+28], ecx :00428406 8BC8 mov ecx, eax :00428408 81E101000080 and ecx, 80000001 :0042840E 7905 jns 00428415 :00428410 49 dec ecx :00428411 83C9FE or ecx, -00000002 :00428414 41 inc ecx --------- :00428415 99 cdq :00428416 2BC2 sub eax, edx :00428418 894C2424 mov dword[esp+24], ecx :0042841C 8BC8 mov ecx, eax :0042841E D1F9 sar ecx, 1 :00428420 B856555555 mov eax, 55555556 :00428425 F7E9 imul ecx :00428427 8BC2 mov eax, edx :00428429 C1E81F shr eax, 1F :0042842C 03C2 add eax, edx :0042842E 8D1440 lea edx, dword[eax+2*eax] :00428431 8BF0 mov esi, eax :00428433 2BCA sub ecx, edx :00428435 B856555555 mov eax, 55555556 :0042843A F7EE imul esi :0042843C 894C241C mov dword[esp+1C], ecx :00428440 8BCA mov ecx, edx :00428442 C1E91F shr ecx, 1F :00428445 03CA add ecx, edx :00428447 8D0449 lea eax, dword[ecx+2*ecx] :0042844A 2BF0 sub esi, eax :0042844C B8398EE338 mov eax, 38E38E39 :00428451 F7E9 imul ecx :00428453 D1FA sar edx, 1 :00428455 89742414 mov dword[esp+14], esi :00428459 8BF2 mov esi, edx :0042845B C1EE1F shr esi, 1F :0042845E 03F2 add esi, edx :00428460 8D14F6 lea edx, dword[esi+8*esi] :00428463 2BCA sub ecx, edx :00428465 B8398EE338 mov eax, 38E38E39 :0042846A F7EE imul esi :0042846C D1FA sar edx, 1 :0042846E 8BC2 mov eax, edx :00428470 C1E81F shr eax, 1F :00428473 03C2 add eax, edx :00428475 8B542410 mov edx, dword[esp+10] :00428479 8D04C0 lea eax, dword[eax+8*eax] :0042847C 2BF0 sub esi, eax :0042847E 8B442424 mov eax, dword[esp+24] :00428482 03442428 add eax, dword[esp+28] :00428486 53 push ebx :00428487 52 push edx :00428488 50 push eax :00428489 89442438 mov dword[esp+38], eax :0042848D 8B442420 mov eax, dword[esp+20] :00428491 50 push eax :00428492 8B442428 mov eax, dword[esp+28] :00428496 89742440 mov dword[esp+40], esi :0042849A 8B542440 mov edx, dword[esp+40] :0042849E 8B742444 mov esi, dword[esp+44] :004284A2 52 push edx :004284A3 8B542430 mov edx, dword[esp+30] :004284A7 55 push ebp :004284A8 50 push eax :004284A9 03F7 add esi, edi :004284AB 56 push esi :004284AC 52 push edx :004284AD 51 push ecx :004284AE 894C2460 mov dword[esp+60], ecx :004284B2 E899FCFFFF call 00428150[/code]Code: Select all
:00425B49 8BD8 mov ebx, eax :00425B4B 81E301000080 and ebx, 80000001 :00425B51 56 push esi :00425B52 57 push edi :00425B53 7905 jns 00425B5A :00425B55 4B dec ebx :00425B56 83CBFE or ebx, -00000002 :00425B59 43 inc ebx --------- :00425B5A 99 cdq :00425B5B 2BC2 sub eax, edx :00425B5D D1F8 sar eax, 1 :00425B5F 8BF0 mov esi, eax :00425B61 81E601000080 and esi, 80000001 :00425B67 7905 jns 00425B6E :00425B69 4E dec esi :00425B6A 83CEFE or esi, -00000002 :00425B6D 46 inc esi --------- :00425B6E 99 cdq :00425B6F 2BC2 sub eax, edx :00425B71 8BC8 mov ecx, eax :00425B73 D1F9 sar ecx, 1 :00425B75 B856555555 mov eax, 55555556 :00425B7A F7E9 imul ecx :00425B7C 8BC2 mov eax, edx :00425B7E C1E81F shr eax, 1F :00425B81 03C2 add eax, edx :00425B83 8D1440 lea edx, dword[eax+2*eax] :00425B86 2BCA sub ecx, edx :00425B88 894C2418 mov dword[esp+18], ecx :00425B8C 8BC8 mov ecx, eax :00425B8E B856555555 mov eax, 55555556 :00425B93 F7E9 imul ecx :00425B95 8BC2 mov eax, edx :00425B97 C1E81F shr eax, 1F :00425B9A 03C2 add eax, edx :00425B9C 8D1440 lea edx, dword[eax+2*eax] :00425B9F 2BCA sub ecx, edx :00425BA1 8BE8 mov ebp, eax :00425BA3 81E501000080 and ebp, 80000001 :00425BA9 894C2414 mov dword[esp+14], ecx :00425BAD 7905 jns 00425BB4 :00425BAF 4D dec ebp :00425BB0 83CDFE or ebp, -00000002 :00425BB3 45 inc ebp --------- :00425BB4 99 cdq :00425BB5 2BC2 sub eax, edx :00425BB7 D1F8 sar eax, 1 :00425BB9 8BC8 mov ecx, eax :00425BBB 81E101000080 and ecx, 80000001 :00425BC1 7905 jns 00425BC8 :00425BC3 49 dec ecx :00425BC4 83C9FE or ecx, -00000002 :00425BC7 41 inc ecx --------- :00425BC8 99 cdq :00425BC9 2BC2 sub eax, edx :00425BCB D1F8 sar eax, 1 :00425BCD 894C2438 mov dword[esp+38], ecx :00425BD1 8BC8 mov ecx, eax :00425BD3 81E101000080 and ecx, 80000001 :00425BD9 7905 jns 00425BE0 :00425BDB 49 dec ecx :00425BDC 83C9FE or ecx, -00000002 :00425BDF 41 inc ecx --------- :00425BE0 99 cdq :00425BE1 2BC2 sub eax, edx :00425BE3 D1F8 sar eax, 1 :00425BE5 894C242C mov dword[esp+2C], ecx :00425BE9 8BC8 mov ecx, eax :00425BEB 81E101000080 and ecx, 80000001 :00425BF1 7905 jns 00425BF8 :00425BF3 49 dec ecx :00425BF4 83C9FE or ecx, -00000002 :00425BF7 41 inc ecx --------- :00425BF8 99 cdq :00425BF9 2BC2 sub eax, edx :00425BFB 894C2428 mov dword[esp+28], ecx :00425BFF 8BC8 mov ecx, eax :00425C01 D1F9 sar ecx, 1 :00425C03 B856555555 mov eax, 55555556 :00425C08 F7E9 imul ecx :00425C0A 8BC2 mov eax, edx :00425C0C C1E81F shr eax, 1F :00425C0F 03C2 add eax, edx :00425C11 8D1440 lea edx, dword[eax+2*eax] :00425C14 8BF8 mov edi, eax :00425C16 2BCA sub ecx, edx :00425C18 B856555555 mov eax, 55555556 :00425C1D F7EF imul edi :00425C1F 894C2420 mov dword[esp+20], ecx :00425C23 8BCA mov ecx, edx :00425C25 C1E91F shr ecx, 1F :00425C28 03CA add ecx, edx :00425C2A 8D0449 lea eax, dword[ecx+2*ecx] :00425C2D 2BF8 sub edi, eax :00425C2F B8398EE338 mov eax, 38E38E39 :00425C34 F7E9 imul ecx :00425C36 D1FA sar edx, 1 :00425C38 897C241C mov dword[esp+1C], edi :00425C3C 8BFA mov edi, edx :00425C3E C1EF1F shr edi, 1F :00425C41 03FA add edi, edx :00425C43 8D14FF lea edx, dword[edi+8*edi] :00425C46 2BCA sub ecx, edx :00425C48 B8398EE338 mov eax, 38E38E39 :00425C4D F7EF imul edi :00425C4F D1FA sar edx, 1 :00425C51 8BC2 mov eax, edx :00425C53 C1E81F shr eax, 1F :00425C56 03C2 add eax, edx :00425C58 8B542438 mov edx, dword[esp+38] :00425C5C 8D04C0 lea eax, dword[eax+8*eax] :00425C5F 2BF8 sub edi, eax :00425C61 8B442428 mov eax, dword[esp+28] :00425C65 56 push esi :00425C66 897C2438 mov dword[esp+38], edi :00425C6A 8D3C2A lea edi, dword[edx+ebp] :00425C6D 8B542430 mov edx, dword[esp+30] :00425C71 03C2 add eax, edx :00425C73 8B542418 mov edx, dword[esp+18] :00425C77 52 push edx :00425C78 8B54243C mov edx, dword[esp+3C] :00425C7C 50 push eax :00425C7D 8944243C mov dword[esp+3C], eax :00425C81 8B442428 mov eax, dword[esp+28] :00425C85 50 push eax :00425C86 8B442428 mov eax, dword[esp+28] :00425C8A 52 push edx :00425C8B 8B542434 mov edx, dword[esp+34] :00425C8F 53 push ebx :00425C90 50 push eax :00425C91 57 push edi :00425C92 52 push edx :00425C93 51 push ecx :00425C94 894C2464 mov dword[esp+64], ecx :00425C98 E8A3FCFFFF call 00425940
The actual response is here:SzG wrote:I don't understand that. There is execution comparison above, I can see no tables.Adam Hair wrote: I believe Robert's response to the disassembled code above was this:
http://forum.computerschach.de/cgi-bin/ ... l?tid=3306
>Robert Houdart wrote:The basic positional values (piece values on squares) are not very different from the Ippo* family. Looking at Ippo's source code, I was - and >still am - amazed at how such simple functions can produce such an extremely strong engine. So if you find relatively similar positional evaluations in some quiet >positions, I will not be surprised.
>The material balance table is part of the "basic positional values", so the infamous thread just confirms what I wrote.
>Robert
Also, some other responses concerning the origins of Houdini:
http://www.open-chess.org/viewtopic.php ... 4&start=20
Robert has maintained that he solely took ideas from the Ippolit family.
If that is indeed true, then there is no crime. Given the degree of similarity
that I have seen between Houdini and Robbolito/IvanHoe, it does seem
that Houdini could properly be called a derivative of Ippolit ( as far as I
understand the use of that term ), which is a statement that I have never
seen Robert make. To be honest, I have never seen him deny it either.
Robert Houdart wrote:Lance Perkins showed a piece of code that is part of the initialization of the material balance table, which is a static table containing the values of different piece configurations. Let me copy here what I already wrote on Talkchess in the very first thread about Houdini 15 days before Lance Perkins' topic.
>Robert Houdart wrote:The basic positional values (piece values on squares) are not very different from the Ippo* family. Looking at Ippo's source code, I was - and >still am - amazed at how such simple functions can produce such an extremely strong engine. So if you find relatively similar positional evaluations in some quiet >positions, I will not be surprised.
The material balance table is part of the "basic positional values", so the infamous thread just confirms what I wrote.
Robert
Thank you for the links, which I hadn't seen. I don't think that there is any question of a "crime" being committed in this particular case, merely an apparent inconsistency and a possible problem for testers.Adam Hair wrote: I believe Robert's response to the disassembled code above was this:
http://forum.computerschach.de/cgi-bin/ ... l?tid=3306
>Robert Houdart wrote:The basic positional values (piece values on squares) are not very different from the Ippo* family. Looking at Ippo's source code, I was - and >still am - amazed at how such simple functions can produce such an extremely strong engine. So if you find relatively similar positional evaluations in some quiet >positions, I will not be surprised.
>The material balance table is part of the "basic positional values", so the infamous thread just confirms what I wrote.
>Robert
Also, some other responses concerning the origins of Houdini:
http://www.open-chess.org/viewtopic.php ... 4&start=20
Robert has maintained that he solely took ideas from the Ippolit family.
If that is indeed true, then there is no crime. Given the degree of similarity
that I have seen between Houdini and Robbolito/IvanHoe, it does seem
that Houdini could properly be called a derivative of Ippolit ( as far as I
understand the use of that term ), which is a statement that I have never
seen Robert make. To be honest, I have never seen him deny it either.
It would seem to be the logical decision. I will take this opportunity to say that I thought that the pressure on you to do so was extremely irritating. I have written complaining about it before and I actually started another post about it this morning, before abandoning it in disgust.Martin Thoresen wrote:CEGT has decided to include Houdini.
http://www.talkchess.com/forum/viewtopic.php?t=37880
Best,
Martin
That's because you know nothing about chess programming.SzG wrote:I don't understand that. There is execution comparison above, I can see no tables.