Looks like Deep Sjeng wins cct12
Moderator: Ras
-
- Posts: 3721
- Joined: Thu Mar 16, 2006 7:10 pm
Re: Looks like Deep Sjeng wins cct12
I think some are using the fact that DS used a cluster almost as if it was somehow cheating. It takes a lot of hard work to get a cluster to work and just because others were not able to doesn't mean that the people who were able to did anything wrong. In fact I think they should be encouraged as that is how the envelope will be pushed. DS was always a strong engine and now it seems even stronger. Some great games even though Rybka was unfortunately absent. I think Komodo showed tremendous potential even though looking at the score of the Junior game it seems that it has a serious bug overlooking several mates in 1. These bugs are to be expected with a work in progress and I think considering that Komodo was playing on a single core, it has shown how evaluation and knowledge based engine can be a great equalizer especially at non blitz time controls. An MP Komodo will be a very interesting prospect indeed.
-
- Posts: 1286
- Joined: Tue Aug 18, 2009 3:06 am
Re: Looks like Deep Sjeng wins cct12
I agree 100%!M ANSARI wrote:I think some are using the fact that DS used a cluster almost as if it was somehow cheating. It takes a lot of hard work to get a cluster to work and just because others were not able to doesn't mean that the people who were able to did anything wrong. In fact I think they should be encouraged as that is how the envelope will be pushed. DS was always a strong engine and now it seems even stronger. Some great games even though Rybka was unfortunately absent. I think Komodo showed tremendous potential even though looking at the score of the Junior game it seems that it has a serious bug overlooking several mates in 1. These bugs are to be expected with a work in progress and I think considering that Komodo was playing on a single core, it has shown how evaluation and knowledge based engine can be a great equalizer especially at non blitz time controls. An MP Komodo will be a very interesting prospect indeed.
I noticed even on the Hiarcs site, I quote Harvey Williamson,
" Looks like we will come 2nd to the Sjeng Cluster on 80-120 cores. " ???
Perhaps, I am reading this wrong, but it seems even a Hiarcs team member is using this same fact as an excuse. Again to me its simple, it one piece of software/engine can use a cluster and it wins, then bully for it. If you can use the better hardware, and win, then it seems some would conclude this is better software/engine. At least for a open hardware tournement like this.
Before anyone thinks I am bashing Hiarcs, stop! I have owned every single Hiarcs engine that has been released, and will continue to do so. Infact, at one time it was by favourite engine, perhaps due to the Dean Hergott and Hiarcs 6.0 match, and many games of it versus Bryon Nickoloff.
-
- Posts: 5106
- Joined: Tue Apr 29, 2008 4:27 pm
Re: Looks like Deep Sjeng wins cct12
Congradulation to Deep Sjeng, it coasted along with no trouble whatsoever, a very strong program!Robert Flesher wrote:Deep Sjeng has an advantage against Crafty and is already a full point ahead of the rest. A draw or win is clear first, whereas even a miracle draw ties for first place.
Very nice!
Don
-
- Posts: 736
- Joined: Wed Dec 02, 2009 9:48 am
Re: Looks like Deep Sjeng wins cct12
M ANSARI wrote:I think some are using the fact that DS used a cluster almost as if it was somehow cheating. It takes a lot of hard work to get a cluster to work and just because others were not able to doesn't mean that the people who were able to did anything wrong. In fact I think they should be encouraged as that is how the envelope will be pushed. DS was always a strong engine and now it seems even stronger. Some great games even though Rybka was unfortunately absent. I think Komodo showed tremendous potential even though looking at the score of the Junior game it seems that it has a serious bug overlooking several mates in 1. These bugs are to be expected with a work in progress and I think considering that Komodo was playing on a single core, it has shown how evaluation and knowledge based engine can be a great equalizer especially at non blitz time controls. An MP Komodo will be a very interesting prospect indeed.
I don't think it was cheating at all because those were the rules. Although I must admit I like tournaments where the hardware is the same whether it be everyone has clusters or everyone has Commodore 64s! Those types of tournaments have a little more of an old mano y mano vibe to me. But I am very excited about the prospects of a new Hiarcs and Deep Sjeng. Especially Hiarcs as it often has unique suggestions during analysis and is a fun playing partner.
I wish Naum 4.1 had entered.
-
- Posts: 5106
- Joined: Tue Apr 29, 2008 4:27 pm
Re: Looks like Deep Sjeng wins cct12
Deep Junior had the bug, in the final position (which was played under protest) Deep Junior could not beat Komodo with 2 rooks vs a king - a common bug usually due to some hash table glitch or multi-processing bug.M ANSARI wrote:I think some are using the fact that DS used a cluster almost as if it was somehow cheating. It takes a lot of hard work to get a cluster to work and just because others were not able to doesn't mean that the people who were able to did anything wrong. In fact I think they should be encouraged as that is how the envelope will be pushed. DS was always a strong engine and now it seems even stronger. Some great games even though Rybka was unfortunately absent. I think Komodo showed tremendous potential even though looking at the score of the Junior game it seems that it has a serious bug overlooking several mates in 1. These bugs are to be expected with a work in progress and I think considering that Komodo was playing on a single core, it has shown how evaluation and knowledge based engine can be a great equalizer especially at non blitz time controls. An MP Komodo will be a very interesting prospect indeed.
The tournament exposed (in my view) that Komodo still needs some work on king safety.
-
- Posts: 5106
- Joined: Tue Apr 29, 2008 4:27 pm
Re: Looks like Deep Sjeng wins cct12
I agree, the rules did not stipulate only singe core engines so I don't see the problem. This was a tournament of computing systems which include both hardware and software and Sjeng wins fair and square.tomgdrums wrote:M ANSARI wrote:I think some are using the fact that DS used a cluster almost as if it was somehow cheating. It takes a lot of hard work to get a cluster to work and just because others were not able to doesn't mean that the people who were able to did anything wrong. In fact I think they should be encouraged as that is how the envelope will be pushed. DS was always a strong engine and now it seems even stronger. Some great games even though Rybka was unfortunately absent. I think Komodo showed tremendous potential even though looking at the score of the Junior game it seems that it has a serious bug overlooking several mates in 1. These bugs are to be expected with a work in progress and I think considering that Komodo was playing on a single core, it has shown how evaluation and knowledge based engine can be a great equalizer especially at non blitz time controls. An MP Komodo will be a very interesting prospect indeed.
I don't think it was cheating at all because those were the rules. Although I must admit I like tournaments where the hardware is the same whether it be everyone has clusters or everyone has Commodore 64s! Those types of tournaments have a little more of an old mano y mano vibe to me. But I am very excited about the prospects of a new Hiarcs and Deep Sjeng. Especially Hiarcs as it often has unique suggestions during analysis and is a fun playing partner.
I wish Naum 4.1 had entered.
-
- Posts: 44318
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
Re: Looks like Deep Sjeng wins cct12
Congratulations to Gian-Carlo for Sjeng's fine victory.
Thanks to Peter for running the tournament under sometimes difficult circumstances.
Thanks to all of those who participated for providing us with a fine spectacle.
Thanks to Peter for running the tournament under sometimes difficult circumstances.
Thanks to all of those who participated for providing us with a fine spectacle.

gbanksnz at gmail.com
-
- Posts: 2006
- Joined: Thu Sep 18, 2008 10:24 pm
Re: Looks like Deep Sjeng wins cct12
Was Komodo on multi-core?
-
- Posts: 5106
- Joined: Tue Apr 29, 2008 4:27 pm
Re: Looks like Deep Sjeng wins cct12
Komodo was on single core. I'm satisfied with the performance considering how much out-gunned we were and the level of the competition, we were playing some very fine programs on some fine hardware!Werewolf wrote:Was Komodo on multi-core?
I hope to have an MP version in April or May - no promises but I want this for the Dutch Computer Chess Championship which we are going to play in.
I hope other programmer can come to this - I plan to be there and so will Larry Kaufman. So it's my hope to have a stronger program AND have an MP version for this tournament.
Last edited by Don on Sun Feb 21, 2010 6:49 pm, edited 1 time in total.
-
- Posts: 5106
- Joined: Tue Apr 29, 2008 4:27 pm
Re: Looks like Deep Sjeng wins cct12
The TD did a fine job. Being a TD is not always a pleasant task.Graham Banks wrote:Congratulations to Gian-Carlo for Sjeng's fine victory.
Thanks to Peter for running the tournament under sometimes difficult circumstances.
Thanks to all of those who participated for providing us with a fine spectacle.