Tony Thomas wrote:The final game was against ZCT, Zach told me that he was going to smoke me, I did expect Romi to win in this game because Romi is much stronger than ZCT on a single core. However, having hardware about 4 times slower, and ZCT being able to finally use 2 cores reduced my chances a lot. Also, Romi doesnt have much clue about endgame. Towards the end, it was Romi (NN) vs ZCT (2pawns, and a bishop), Romi's knights were just dancing around instead of trying to capture those pawns..and Romi didnt realize until it was too late that she was losing. Oh well, may be next time.
Haha, of course you realize I was playing about the whole smoking thing. I think it was also luck on ZCT's side, anything can happen in one game. Romi was searching pretty deeply considering the hardware disadvantage, but I think ZCT having marginally more passed pawn eval helped it win. I see you also stole my dancing analogy. Nice game though, very entertaining.
I'm writing up my tournament report now, I should probably finish it tomorrow.
Sorry for using your analogy..I forgot that dancing was your thing..Disco didnt die, Zach Wagner killed it.. How long is your report going to be?
Romi's results have to be filtered by the 'Celeron' filter. Yes Romi's use of time was sub optimal for the hardware difference. And Romi would have done much better with a smoother time usage. However, Romi's time uage was designed first for computer -vs- human play and second for computer -vs- computer play using equal hardware.
Even though, under the circumstances of this event Romi's time usage seemed buggy, in reality if Romi was running on a machine that got 2x nodes what the Celeron got then Romi would have been able to do more in the last two minutes and would have had better positions from which to do it.
Note, the 50 elo for doubleing the node rate is a myth it is much larger than that. Take WChess running on a 90 MHz pentium. In the mid 90's(?) it was smashing grandmasters at Harvard(?) and at Ageon(?) and had a rating of about 2550(?). If WChess was playing in this tournament and still running on a 90 MHz pentium it would be lucky to get a 1600 elo performance. And Rybka running on todays hardware back in the mid 90's would be rated 4000+ rather than 3000+.
If you are on a sidewalk and the covid goes beep beep
Just step aside or you might have a bit of heat
Covid covid runs through the town all day
Can the people ever change their ways
Sherwin the covid's after you
Sherwin if it catches you you're through
Romi's time management would work well against weaker engines as searching longer earlier in the game would give Romi significant advanage against them. That's not the case against stronger engines, yes Romi would have an equal position against them, but with those few seconds left on the clock Romi would simply get out serached. I suggest using a more conservative algorithm for time management, something like the one used in Crafty or Fruit. Yea, I am looking at what I can do about the hardware, but I need a Romi that wont get flagged. In about 10 games against Ermintrude the other day, Romi got flagged 3 times and won the other 7, getting flagged in won position didnt help Romi's rating any.
Romi searched around 14-18 plies on my crappy hardware during middle game regards...
But, then it was not Romi messing up. It was caused by using polyglot. Tony was just correcting what went wrong with polyglot.
Hey Tony, best to just use winboard by itself and forget about a book.
If there is a complaint against what was done then we will have to bite the bullet.
Prophet is my engine, and I have no complaint. Best of luck to Romi in the remainder of the tournament.
--
James
James, did you analyze the game? Is it possible for you to post the game for everyone to see? I am pretty sure it was just a blunder from prophet's side.
You want me to relive that??
No - I haven't analyzed it yet, but I will later this week. Prophet's logfile shows +1.8ish, then down to -6 or so in just a few moves. If you want to take a look it's here: http://chessprogramming.org/prophet/acc ... c_2008.php . I'll get the PGN up there tomorrow and try to analyze it later in the week. I was waiting on the PGNs to be posted but I understand from reading another thread that I'll have to get the PGNs myself from ICC.
But, then it was not Romi messing up. It was caused by using polyglot. Tony was just correcting what went wrong with polyglot.
Hey Tony, best to just use winboard by itself and forget about a book.
If there is a complaint against what was done then we will have to bite the bullet.
Prophet is my engine, and I have no complaint. Best of luck to Romi in the remainder of the tournament.
--
James
James, did you analyze the game? Is it possible for you to post the game for everyone to see? I am pretty sure it was just a blunder from prophet's side.
You want me to relive that??
No - I haven't analyzed it yet, but I will later this week. Prophet's logfile shows +1.8ish, then down to -6 or so in just a few moves. If you want to take a look it's here: http://chessprogramming.org/prophet/acc ... c_2008.php . I'll get the PGN up there tomorrow and try to analyze it later in the week. I was waiting on the PGNs to be posted but I understand from reading another thread that I'll have to get the PGNs myself from ICC.
--
James
James you are a good guy, I do not want you to relive that. If you want the PGN for the games, you can go to history and then
"/mailstored Prophetx (game#)", and ICC will send you the game. Romi vs Prophet game is right on the top if you want that. I will let the analyzing specialist dice and examine the game.
Tony Thomas wrote:Romi's time management would work well against weaker engines as searching longer earlier in the game would give Romi significant advanage against them. That's not the case against stronger engines, yes Romi would have an equal position against them, but with those few seconds left on the clock Romi would simply get out serached. I suggest using a more conservative algorithm for time management, something like the one used in Crafty or Fruit. Yea, I am looking at what I can do about the hardware, but I need a Romi that wont get flagged. In about 10 games against Ermintrude the other day, Romi got flagged 3 times and won the other 7, getting flagged in won position didnt help Romi's rating any.
Romi searched around 14-18 plies on my crappy hardware during middle game regards...
Tony
I will do some experimenting at smoothing out the time management.
If you are on a sidewalk and the covid goes beep beep
Just step aside or you might have a bit of heat
Covid covid runs through the town all day
Can the people ever change their ways
Sherwin the covid's after you
Sherwin if it catches you you're through