I get this impression when i see the Mark Thoresen (tcec long timings) matches, that R4 is clearly superior to all else.
I'm begining to wonder if perhaps
1). It IS worthy of 2 years work
2). That Vas did NOT dumb it down.
3). And even that it is clearly more than could ever make sense as a Rybka 3.5 patch.
And that maybe the elo increase from Rybkas 3 to 4, in numbers, may be much more significant, than the higher number of elo points between Rybkas 2 to 3.
Either these thoughts are correct or they're not. I don't know. But I am going by the feeling i get from TCEC matches.
It's hard to believe that anyone testing R4 vs other top engines, under these conditions, will see Rybka lose a match.
Perhaps Rybka 4 IS proving itself.
Moderators: hgm, Rebel, chrisw
-
- Posts: 8514
- Joined: Thu Mar 09, 2006 3:25 am
- Location: Jerusalem Israel
-
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: Perhaps Rybka 4 IS proving itself.
Hi,
so you will have one or two matches only which say nothing. Not a different to have such a match with 40 in 120 or 40 in 1.
Need to many time or a lot of hardware power must here if you like such results.
The main problem in testing. After all my experience with the different time controls I think that different programs can be 30 ELO stronger if I compare to the others.
Possible that Rybka is one of these programs because its play strong in the end of the middle game, the most important playing phase for computer chess engines.
I made for some months on my Q6600 a little experiment, ponder = on, 64, 128, 256, 512 hash, 4-pieces ...
SF 1.6.3 vs. Rybka 3, both 32bit version, xp pro, Shredder GUI, Shredder 12 opening book. Both engines = 1 Core.
40 in 3
18,5 : 31,5, 64Mb Hash
40 in 5
17,5 : 32,5, 128Mb Hash
40 in 10
22,0 : 28,0, 256Mb Hash
40 in 20
23,5 : 26,5, 256Mb Hash
40 in 40
24,0 : 26,0, 512Mb Hash
And now?
Nothing, interesting to see but not more or less.
It's only one match which say nothing!
To many random in 50 games only.
But with a lot fantasy you can say ... with 40 in 120 SF win the match. All is possible ... maximal 30 ElO more or less I think is the different between time controls, not speaking from 1+1 ... speaking from 40 in 10 up to 40 in 600 ... think so, not sure but after all information I collect.
Best
Frank
so you will have one or two matches only which say nothing. Not a different to have such a match with 40 in 120 or 40 in 1.
Need to many time or a lot of hardware power must here if you like such results.
The main problem in testing. After all my experience with the different time controls I think that different programs can be 30 ELO stronger if I compare to the others.
Possible that Rybka is one of these programs because its play strong in the end of the middle game, the most important playing phase for computer chess engines.
I made for some months on my Q6600 a little experiment, ponder = on, 64, 128, 256, 512 hash, 4-pieces ...
SF 1.6.3 vs. Rybka 3, both 32bit version, xp pro, Shredder GUI, Shredder 12 opening book. Both engines = 1 Core.
40 in 3
18,5 : 31,5, 64Mb Hash
40 in 5
17,5 : 32,5, 128Mb Hash
40 in 10
22,0 : 28,0, 256Mb Hash
40 in 20
23,5 : 26,5, 256Mb Hash
40 in 40
24,0 : 26,0, 512Mb Hash
And now?
Nothing, interesting to see but not more or less.
It's only one match which say nothing!
To many random in 50 games only.
But with a lot fantasy you can say ... with 40 in 120 SF win the match. All is possible ... maximal 30 ElO more or less I think is the different between time controls, not speaking from 1+1 ... speaking from 40 in 10 up to 40 in 600 ... think so, not sure but after all information I collect.
Best
Frank
-
- Posts: 2041
- Joined: Wed Mar 08, 2006 8:30 pm
Re: Perhaps Rybka 4 IS proving itself.
Good grief, Frank, here you are doing it again: 50-game matches...Frank Quisinsky wrote: 18,5 : 31,5,
17,5 : 32,5,
22,0 : 28,0,
....
Why don't you take a Statistics 101 course?
... and all this, only to answer some guy's "impression"
-
- Posts: 1833
- Joined: Thu Jun 22, 2006 12:07 am
Re: Perhaps Rybka 4 IS proving itself.
First of all, my name is Martin, not Mark.S.Taylor wrote:I get this impression when i see the Mark Thoresen (tcec long timings) matches, that R4 is clearly superior to all else.
I'm begining to wonder if perhaps
1). It IS worthy of 2 years work
2). That Vas did NOT dumb it down.
3). And even that it is clearly more than could ever make sense as a Rybka 3.5 patch.
1: I don't agree, I still feel that R4 should have been R3+.
2: Impossible for me to answer but my gut feeling is that he did dumb it down, perhaps only slightly.
3: See #1.
-
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: Perhaps Rybka 4 IS proving itself.
Hello Ernest,
I don't know what you mean!
If you like, you can write in German language.
Surely I will give you an reply in "my" English and German.
Have a nice weekend!
Best
Frank
I don't know what you mean!
If you like, you can write in German language.
Surely I will give you an reply in "my" English and German.
Have a nice weekend!
Best
Frank
-
- Posts: 1777
- Joined: Thu Jun 05, 2008 10:58 pm
- Location: Canada
Re: Perhaps Rybka 4 IS proving itself.
I have been testing Deep Rybka vs. Deep Shredder at very long time controls. Many of the games are draws but of the ones with a win it is about 5 to 1 in favour of Rybka. Rybka is a true powerhouse.S.Taylor wrote: It's hard to believe that anyone testing R4 vs other top engines, under these conditions, will see Rybka lose a match.
I was disappointed at Shredder's poor showing at the WCCC this year. Recall the days when Shredder would normally win that tourney.
Cordially,
Sean
-
- Posts: 3707
- Joined: Thu Mar 16, 2006 7:10 pm
Re: Perhaps Rybka 4 IS proving itself.
Rybka 4 is probably the most underrated engine out. During beta testing it was obvious that this engine could play one hell of a mean chess game. But it was hampered with an incredibly terrible time control mechanism. We tried very hard to sort that out in the limited time, but I think it was a band aid on a very large wound. Rybka 4 probably has one of the worst time management algo's out there, and if that would be improved it would dramatically improve in the ELO rating scheme. At long time controls, the weakness of time management is somewhat covered over (although not eliminated) and I saw that during beta testing. Hopefully this issue and a few other issues will be included in the bug fix everyone has been waiting for. My guess for the poor time management is that it is pulled off directly from the cluster version. That was designed to incorporate latencies to allow for remote play and Slave-Master LAN communication, where time is really not so critical. Move that system to a normal PC setup at fast time controls and those latencies will just kill you.
-
- Posts: 6340
- Joined: Mon Mar 13, 2006 2:34 pm
- Location: Acworth, GA
Re: Perhaps Rybka 4 IS proving itself.
Yes, I agree with you on this very much. I noticed this from day one of the engines release when testing the engine on playchess.M ANSARI wrote:Rybka 4 is probably the most underrated engine out. During beta testing it was obvious that this engine could play one hell of a mean chess game. But it was hampered with an incredibly terrible time control mechanism. We tried very hard to sort that out in the limited time, but I think it was a band aid on a very large wound. Rybka 4 probably has one of the worst time management algo's out there, and if that would be improved it would dramatically improve in the ELO rating scheme. At long time controls, the weakness of time management is somewhat covered over (although not eliminated) and I saw that during beta testing. Hopefully this issue and a few other issues will be included in the bug fix everyone has been waiting for. My guess for the poor time management is that it is pulled off directly from the cluster version. That was designed to incorporate latencies to allow for remote play and Slave-Master LAN communication, where time is really not so critical. Move that system to a normal PC setup at fast time controls and those latencies will just kill you.
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
__________________________________________________________________
Ted Summers
-
- Posts: 8514
- Joined: Thu Mar 09, 2006 3:25 am
- Location: Jerusalem Israel
Re: Perhaps Rybka 4 IS proving itself.
Martin!Martin Thoresen wrote:First of all, my name is Martin, not Mark.S.Taylor wrote:I get this impression when i see the Mark Thoresen (tcec long timings) matches, that R4 is clearly superior to all else.
I'm begining to wonder if perhaps
1). It IS worthy of 2 years work
2). That Vas did NOT dumb it down.
3). And even that it is clearly more than could ever make sense as a Rybka 3.5 patch.
1: I don't agree, I still feel that R4 should have been R3+.
2: Impossible for me to answer but my gut feeling is that he did dumb it down, perhaps only slightly.
3: See #1.
Very sorry for such a terrible mistake!
Yes, there still IS much more yet to improve. But maybe it's 50-100 elo increase ove Rybka 3, and, maybe it takes much more work to increase the elo, than it did since Rybka 3 level had been reached.
But still, maybe I'm wrong, and Vas was only taking advantage of the lack of competition, as well as knowing that the clones would never reach Rybka 4.
But I was also being devil's advocate, to know what people think now, as i myself cannot deny that from your (Martin) matches, Rybka is still supreme.
I am very much looking forward to Decemeber/christmas time, when we will see if there is a change of number 1. If there isn't, then that would seem to prove Rybka 4 rather well, I think. I don't care if it is a clone derivative. Let it just be seen (obviously without a new wave of Ippolits)!
-
- Posts: 6401
- Joined: Thu Mar 09, 2006 8:30 pm
- Location: Chicago, Illinois, USA
Re: Perhaps Rybka 4 IS proving itself.
Could you describe the symptoms for people who do not have it?M ANSARI wrote:Rybka 4 is probably the most underrated engine out. During beta testing it was obvious that this engine could play one hell of a mean chess game. But it was hampered with an incredibly terrible time control mechanism. We tried very hard to sort that out in the limited time, but I think it was a band aid on a very large wound. Rybka 4 probably has one of the worst time management algo's out there, and if that would be improved it would dramatically improve in the ELO rating scheme. At long time controls, the weakness of time management is somewhat covered over (although not eliminated) and I saw that during beta testing. Hopefully this issue and a few other issues will be included in the bug fix everyone has been waiting for. My guess for the poor time management is that it is pulled off directly from the cluster version. That was designed to incorporate latencies to allow for remote play and Slave-Master LAN communication, where time is really not so critical. Move that system to a normal PC setup at fast time controls and those latencies will just kill you.
Loses on time? rushes it too much?
Miguel