Which IvanHoe Build Is Best?
- kingliveson
- Posts: 1388
- Joined: Thu Jun 10, 2010 1:22 am
- Real Name: Franklin Titus
- Location: 28°32'1"N 81°22'33"W
Which IvanHoe Build Is Best?
Which IvanHoe build is better? This is not really a topic of interest, but because we have a certain character (and he is not alone) who has taken it upon himself to interject unprompted at each turn on a topic for which he has limited knowledge, this thread is warranted. It's not every idol or foolish talk that demands a reply, but lest a false notion repeated frequently take hold, it is however, occasionally necessary that one provides corrections.
Reading some of the comments that were pointed out to me, evoked laughter because you don't expect anyone to take them seriously. Here was one; "Only problem is remember KLO's compiles are approx. 30 to 70 elo weaker than PPs in Windows." KLO (king-lives-on) of course are my builds, and PP, are those of Peterpan's. This is coming from someone, who if you read some of his postings, you'd wonder how this individual turns on a computer without the help of technical-support.
Unless the source code has been modified and inadvertently weakened, the builds should not be more than +/- ~5 Elo apart -- and there are some good supporting data. Frank Quisinsky, who hosted the now closed SWCR rating list, and another talkchess user by the name of Gerhard Sonnabend, both ran well controlled tournaments using IvanHoe 999946f and 999946h respectively, that contradict misconceptions put out there.
Rank Name Elo + - games score oppo. draws
1 Houdini 2.0c x64 3018 19 19 1288 81% 2764 26%
2 Houdini 1.5 x64 2998 14 14 2320 78% 2771 29%
3 Houdini 1.5 w32 2980 19 19 1240 79% 2756 29%
4 Rybka 4 x64 Exp. 42 2966 19 18 1280 78% 2743 28%
5 Rybka 4.1 x64 Exp. 79TD v.1 2963 20 19 1200 79% 2735 26%
6 IvanHoe 999946f x64 2962 17 17 1400 75% 2770 33%
7 Komodo 3.0 x64 2961 16 16 1568 75% 2762 33%
8 Rybka 4 x64 Exp. 61 2959 21 21 1000 78% 2741 27%
9 IvanHoe B46fB x64 2956 17 17 1400 74% 2770 35%
10 Critter 1.2 x64 2956 15 15 1768 75% 2760 34%
http://talkchess.com/forum/viewtopic.ph ... 573#438573
3. Critter 1.4 x64 1CPU ELO 3189 / 1300 games
4. Rybka 4.0 x64 1CPU ELO 3172 / 6142 games
-> IvanHoe 46bh (KLO) x64 1CPU ELO 3167 / 700 games
-> IvanHoe 46h (PP) x64 1CPU ELO 3165 / 700 games
5. Komodo 4.0 x64 1CPU ELO 3163 / 1400 games
6. Stockfish 2.1 x64 1CPU ELO 3142 / 3142 games
http://computerschach.forumfrei.net/t23 ... itions#392
On Frank's test, my build came out 6 Elo ahead, and while on Gerhard's, 2 Elo ahead.
Furthermore, when IvanHoe 999946 (999946a) series came out, I was the only individual who put out the complete build. There was a reason why neither Peterpan nor anyone else was successful. In fact, the library included in my release had to be borrowed. None of these uninformed characters understand, and with 100% certainty, I am very sure neither does Peterpan. The point is, I am quite familiar with the source code and know what am doing.
This is not an attempt to be disparaging, but it was some of these same set of characters who goaded Izak (Peterpan), to ask Critter's author for his source code because they felt the binary provided was not the best that could be produced from source code they've never seen. How comical is that?!
"Remember I told you KLO's were the only Ivanhoe Linux compiles. I never said they were worth a shit."
It is not everything the right hand does, the left hand knows. Anyone who's been following IvanHoe knows some of my contributions. When you are doing someone a favour and don't even ask or care for a thank you, a little courtesy is not too much to be expected.
Reading some of the comments that were pointed out to me, evoked laughter because you don't expect anyone to take them seriously. Here was one; "Only problem is remember KLO's compiles are approx. 30 to 70 elo weaker than PPs in Windows." KLO (king-lives-on) of course are my builds, and PP, are those of Peterpan's. This is coming from someone, who if you read some of his postings, you'd wonder how this individual turns on a computer without the help of technical-support.
Unless the source code has been modified and inadvertently weakened, the builds should not be more than +/- ~5 Elo apart -- and there are some good supporting data. Frank Quisinsky, who hosted the now closed SWCR rating list, and another talkchess user by the name of Gerhard Sonnabend, both ran well controlled tournaments using IvanHoe 999946f and 999946h respectively, that contradict misconceptions put out there.
Rank Name Elo + - games score oppo. draws
1 Houdini 2.0c x64 3018 19 19 1288 81% 2764 26%
2 Houdini 1.5 x64 2998 14 14 2320 78% 2771 29%
3 Houdini 1.5 w32 2980 19 19 1240 79% 2756 29%
4 Rybka 4 x64 Exp. 42 2966 19 18 1280 78% 2743 28%
5 Rybka 4.1 x64 Exp. 79TD v.1 2963 20 19 1200 79% 2735 26%
6 IvanHoe 999946f x64 2962 17 17 1400 75% 2770 33%
7 Komodo 3.0 x64 2961 16 16 1568 75% 2762 33%
8 Rybka 4 x64 Exp. 61 2959 21 21 1000 78% 2741 27%
9 IvanHoe B46fB x64 2956 17 17 1400 74% 2770 35%
10 Critter 1.2 x64 2956 15 15 1768 75% 2760 34%
http://talkchess.com/forum/viewtopic.ph ... 573#438573
3. Critter 1.4 x64 1CPU ELO 3189 / 1300 games
4. Rybka 4.0 x64 1CPU ELO 3172 / 6142 games
-> IvanHoe 46bh (KLO) x64 1CPU ELO 3167 / 700 games
-> IvanHoe 46h (PP) x64 1CPU ELO 3165 / 700 games
5. Komodo 4.0 x64 1CPU ELO 3163 / 1400 games
6. Stockfish 2.1 x64 1CPU ELO 3142 / 3142 games
http://computerschach.forumfrei.net/t23 ... itions#392
On Frank's test, my build came out 6 Elo ahead, and while on Gerhard's, 2 Elo ahead.
Furthermore, when IvanHoe 999946 (999946a) series came out, I was the only individual who put out the complete build. There was a reason why neither Peterpan nor anyone else was successful. In fact, the library included in my release had to be borrowed. None of these uninformed characters understand, and with 100% certainty, I am very sure neither does Peterpan. The point is, I am quite familiar with the source code and know what am doing.
This is not an attempt to be disparaging, but it was some of these same set of characters who goaded Izak (Peterpan), to ask Critter's author for his source code because they felt the binary provided was not the best that could be produced from source code they've never seen. How comical is that?!
"Remember I told you KLO's were the only Ivanhoe Linux compiles. I never said they were worth a shit."
It is not everything the right hand does, the left hand knows. Anyone who's been following IvanHoe knows some of my contributions. When you are doing someone a favour and don't even ask or care for a thank you, a little courtesy is not too much to be expected.
PAWN : Knight >> Bishop >> Rook >>Queen
Re: Which IvanHoe Build Is Best?
I like Ivanhoe B46fC x64 (Peterpan compile)kingliveson wrote:Which IvanHoe build is better?
In my tests (2'+1", 200 games and 164 games), it got 44% against Houdini 2.0c and equality against Critter 1.4
- 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: Which IvanHoe Build Is Best?
Ernest, post your results after you've ran enough games in controlled settings with multiple opponents.
Here is a sample: Played under "CEGT-Blitz-Conditions".
Here is a sample: Played under "CEGT-Blitz-Conditions".
IvanHoe 46bh x64 1CPU (KLO) IvanHoe 46h x64 1CPU (Peterpan) vs ELO Result Perf. ELO Result Perf. Chiron 1.1a x64 1CPU 3025 74.5-25.5 3211 3025 70.0-30.0 3172 Critter 1.4 x64 1CPU 3189 44.0-56.0 3147 3189 43.0-57.0 3140 Houdini 2.0c x64 1CPU 3240 42.5-57.5 3187 3240 36.5-63.5 3144 Komodo 4.0 x64 1CPU 3163 46.0-54.0 3135 3163 49.0-51.0 3156 Loop 2010-x x64 1CPU 3042 66.5-33.5 3161 3042 67.5-32.5 3169 Rybka 4.0 x64 1CPU 3172 51.5-48.5 3182 3172 50.5-49.5 3175 Stockfish 2.1 x64 1CPU 3142 53.5-46.5 3166 3142 60.0-40.0 3212 Ø 3139 378.5-321.5 54.1% 3139 376.5-323.5 53.8% ELO-Stat 54.1% Ø 3139 3167 53.8% Ø 3139 3165
PAWN : Knight >> Bishop >> Rook >>Queen
- 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: Which IvanHoe Build Is Best?
Sweet Mary, Holy Venerated Mother of Baby Jesus, stop the press...stop the press. My IvanHoe build is 28 Elo stronger than the current best. After numerous, more than enough games, both engines using default settings, 128 MB hash, ponder off, tablebases off, 4 cores, 25 opening positions, and 1 minute sudden death time control, a clear undisputed champion emerged. I am thinking this program should now be renamed, packaged, and sold.ernest wrote:I like Ivanhoe B46fC x64 (Peterpan compile)kingliveson wrote:Which IvanHoe build is better?
In my tests (2'+1", 200 games and 164 games), it got 44% against Houdini 2.0c and equality against Critter 1.4
Code: Select all
IvanHoe vs IvanHoe, Blitz 1m 2012
1 IvanHoe 9.46b x64 +28 +11/=32/-7 54.00% 27.0/50
2 Ivanhoe B46fC x64 -28 +7/=32/-11 46.00% 23.0/50
Wait, wait, I am being told there exists 10 more stronger than the best.
Not trying to fight with you []. I would just like you to say which IvanHoe you think is the best, if you have an opinion. Listing 7 "that are better than what I am using", and saying that you could name 4 or 5 more really does not help.
I will send you all the strong PP versions- 64 and 32 bit- if you like. And I have B46fc. But it is nowhere close to the top 6 or 8. It is no slouch- but at least ONE person-you - have seen the light and are headed for some good ones, ie those that don't begin with "9" in the name.
In the words of Don Dailey, "My compiler is better than your compiler."With having different match lengths, I have settled on all "50 game matches" run at 40/3 repeating. All engines like this time limit, and when I wake up after 8 hours sleep, it will in general have run 46 or so of the 50 games.
I am using Rybka 4.1 as a baseline, and comparing each engine I run to that baseline. So the elo difference between 2 engines as it shows in a rating list will be determined by how each performs ag. Rybka. Example: "engine A" shows +70 elo in match with Rybka, and "engine B" shows +35 elo difference. That means in the list "engine A" will list as 35 elo stronger than "engine B".
- Attachments
-
- 20120403_iv_iv_1m_50r.pgn
- (295.65 KiB) Downloaded 561 times
PAWN : Knight >> Bishop >> Rook >>Queen
Re: Which IvanHoe Build Is Best?
Could be, but I'm having fun anyway with my tests!...kingliveson wrote:you are deceiving yourself.
Re: Which IvanHoe Build Is Best?
IvanHoe 9.46 (compiled by KLO)
http://www.chess2u.com/t4670p60-ivanhoe
Thank you so much Kingliveson for this a great version..
Code: Select all
1 Houdini 2.0c x64 3018 19 19 1302 80% 2765 27%
2 Houdini 1.5 x64 2999 14 14 2320 78% 2772 29%
3 Houdini 1.5 w32 2981 19 18 1240 79% 2756 29%
4 Rybka 4 x64 Exp. 42 2967 19 18 1280 78% 2744 28%
5 Rybka 4.1 x64 Exp. 79TD v.1 2963 19 19 1200 79% 2736 26%
6 IvanHoe 999946f x64 2963 17 17 1400 75% 2770 33%
7 Komodo 3.0 x64 2960 16 16 1581 75% 2762 33%
8 Rybka 4 x64 Exp. 61 2959 21 21 1000 78% 2741 27%
9 IvanHoe B46fB x64 2956 17 17 1400 74% 2770 35%
Thank you so much Kingliveson for this a great version..
Re: Which IvanHoe Build Is Best?
+1Marwan wrote:IvanHoe 9.46 (compiled by KLO)
http://www.chess2u.com/t4670p60-ivanhoeCode: Select all
1 Houdini 2.0c x64 3018 19 19 1302 80% 2765 27% 2 Houdini 1.5 x64 2999 14 14 2320 78% 2772 29% 3 Houdini 1.5 w32 2981 19 18 1240 79% 2756 29% 4 Rybka 4 x64 Exp. 42 2967 19 18 1280 78% 2744 28% 5 Rybka 4.1 x64 Exp. 79TD v.1 2963 19 19 1200 79% 2736 26% 6 IvanHoe 999946f x64 2963 17 17 1400 75% 2770 33% 7 Komodo 3.0 x64 2960 16 16 1581 75% 2762 33% 8 Rybka 4 x64 Exp. 61 2959 21 21 1000 78% 2741 27% 9 IvanHoe B46fB x64 2956 17 17 1400 74% 2770 35%
Thank you so much Kingliveson for this a great version..
I concur.
Re: Which IvanHoe Build Is Best?
http://www.chess2u.com/t5680-warning-iv ... lease-readWARNING: Ivanhoe B46fa w32 Contains Trojans and Viruses...
Ivanhoe B46fa: http://depositfiles.com/files/4f0zpef56
Viruscan is reporting that Ivanhoe B46fa w32 contains Trojans and Viruses!
Trojan.Crypt!IK
UnclassifiedMalware
Trojan.Crypt
http://r.virscan.org/report/07237f6d108 ... 5e7ab.html
and virustotal also is reporting that Ivanhoe B46fa w32 contains Trojans and Viruses!
TR/Crypt.XPACK.Gen2
UnclassifiedMalware
Trojan.Crypt!IK
Trojan.Crypt
W32/Obfuscated_J.FP
https://www.virustotal.com/file/368309d ... 333560742/
Trojan.Crypt is a backdoor dropper trojan that injects itself into Windows Win32 system files via IE security holes, and drops various malicious programs onto the infected local or network computer. Win32.Trojan.Crypt can steal and foward passwords, usernames, credit card and other information to outlaying parties. Win32.Trojan.Crypt is serious security threat that can harm both Windows system and PC hardware!
http://www.scanforfree.com/07/remove-wi ... crypt.html
- 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: Which IvanHoe Build Is Best?
http://immortalchess.net/forum/showpost ... count=6405Perterpan wrote: I sometimes wish i wouldn't read other chess forums,but unfortunately i sometimes do
This is what KLO,thinks about me and my work of Ivanhoe
I have to mention,that i have never ever,badmouthed or said anything wrong regarding KLO that i know of.
I also want to mention,that the only reason i "borrowed" KLO's library was because i had the flew and did not have proper time ( i was lazy mostly) to do it that day.A day or 2 later,i can't quite remember,i did put out my own RobboBase Library.
There seems to be a hatred/jealousy conspiracy towards me and my Ivanhoe compiles.
I wish to bring to the attention of all green goblins or haters of me that i am nothing special,i have never written my own chess engine.BUT... i have successfully on my own without the help of others compiled each and every source of Ivanhoe since day 1 until now.I have contributed more bug fixed than anything else besides the main programmers of Ivanhoe.
I feel ashamed and depressed that things has come to this.
But like one of the clever Ippolit Wiki authors once said,well i can't exact quote,but it was something in the view of let the people write what they want on Ippolit wiki and they will give freedom and in so doing each and every persons good/bad becomes visible to everyone.So no need to delete those posts.
In this same spirit i will quote KLO from openchess forum,he posted this on 3 of April.
Just once more,please,please get it through your thick skulls that Ivanhoe DOES NOT BELONG TO me or you,I did not create it,neither did you.
To the creators of Ivanhoe,my hats off,it belongs to you.
Okay here is the quote exact of KLO,you be the judge
http://www.open-chess.org/viewtopic.php?f=3&t=1924
KLO if you are reading this,or whatever other aliases you use,i am bitterly disappointed in you
It appears that Peterpan is emotional disturbed by this thread. Well, I am sorry your feelings are hurt, but this thread was not directed at you. You ought to read it again. It was a jeer at many ignorant statements made by uninformed characters over the course of several months.
No, the reason your compile was unsuccessful was because 999946a publicly released source code was designed for Linux only and had to be ported to Windows. And the reason you were able to compile the next public release (999946e) was because that code had then been ported -- someone had to do the porting. It was not a dig, but rather pointing out an intimate knowledge. Do you understand what it means that it is not everything the right hand does, the left hand knows, or do I have to spell out ABC?
This is a bit shameful; of all people, you, accusing another person of using aliases. You were recently accused of just that, and I came to your defense. How did it make you feel, given you are quite an emotional person, that someone is putting something out there that is blatantly false? If you are going to make such silly accusation, you should accompany it with concrete facts.
I am not sure exactly what is there to be jealous about, the hatred, or who these co-conspirators are, unless you are really starting to believe the hype. Which part of my statement did you find to be false, badmouthing, or offensive? For the record, I think you've done an outstanding job keeping up and building IvanHoe. So let's calm it and take it down a notch; do not be so bitter and depressed, control your emotions, and read the thread again because this was not about you.
PAWN : Knight >> Bishop >> Rook >>Queen
Re: Which IvanHoe Build Is Best?
Code: Select all
Games Completed = 1000 of 1000 (Avg game length = 19.520 sec)
Settings = RR/32MB/500ms+150ms/M 1000cp for 12 moves, D 150 moves/EPD:openings.epd(4000)
1. Houdini 1.5a (64 bits for all) 262.5/400 186-61-153 (L: m=22 t=3 i=0 a=36) (D: r=62 i=33 f=30 s=2 a=26) (tpm=134.4 d=13.2 nps=2645822)
2. IvanHoe 46f (KLO compile) 195.0/400 92-102-206 (L: m=40 t=0 i=0 a=62) (D: r=129 i=33 f=23 s=0 a=21) (tpm=136.7 d=13.7 nps=2158993)
3. IvanHoe .x64.VE.f (VE compile) 146.0/400 50-158-192 (L: m=90 t=0 i=0 a=68) (D: r=130 i=19 f=28 s=1 a=14) (tpm=139.4 d=13.0 nps=1266396)
4. IvanHoe B46fc (PP compile) 206.0/400 110-98-192 (L: m=39 t=0 i=0 a=59) (D: r=126 i=30 f=22 s=1 a=13) (tpm=136.7 d=13.8 nps=2289278)
5. IvanHoe 47c+GH (GH compile) 190.5/400 89-108-203 (L: m=46 t=0 i=0 a=62) (D: r=137 i=25 f=27 s=0 a=14) (tpm=136.5 d=13.6 nps=2246982)
Code: Select all
1 Houdini 1.5a XXXXXXXXXXX / 67.0 - 33.0 / 60.5 - 39.5 / 61.0 - 39.0 / 74.0 - 26.0 262.5/400
2 IvanHoe B46fc (PP compile) 33.0 - 67.0 / XXXXXXXXXXX / 49.0 - 51.0 / 58.5 - 41.5 / 65.5 - 34.5 206.0/400
3 IvanHoe 46f (KLO compile) 39.5 - 60.5 / 51.0 - 49.0 / XXXXXXXXXXX / 50.5 - 49.5 / 54.0 - 46.0 195.0/400
4 IvanHoe 47c+GH (GH compile) 39.0 - 61.0 / 41.5 - 58.5 / 49.5 - 50.5 / XXXXXXXXXXX / 60.5 - 39.5 190.5/400
5 IvanHoe .x64.VE.f (VE compile) 26.0 - 74.0 / 34.5 - 65.5 / 46.0 - 54.0 / 39.5 - 60.5 / XXXXXXXXXXX 146.0/400
- First I want to say I'm not sure having the best compile for PP,VE,KLO or GH ... I just take the latest ones I find.
- Now ... What can I conclude from this test ?
1. No one compile makes at least 40% against Houdini 1.5a
2. KLO compile scores the best against Houdini.
3. KLO compile just beats PP compile by a very little margin.
4. PP compile seems the faster (even faster than GH compile).
5. PP scores the best overall the other Ivanhoe compiles ...
6. VE compile seems a little bit "light" in this test.
7. GH compile even with 47 version instead 46 for others is again a good compile.
If somebody or one of these compilers think having a better compile, please send me the link and I 'll do the same test.