57.2 implies about 50 ELO which is a lot more than the previous version did so I have to say I am skeptical, but seeing is believing so we will have to wait to see.
Why are you skeptical ?
Well, one reason might be that Robert does not have much credibility, because he refuses to acknowledge that Houdini was (is?) 99% copied code from Ippolit. He could still acknowledge this fact and continue to sell his program, but he doesn't, so why should anyone believe him? Anyway, based on my experience with Rybka and Komodo, a gain of 50 elo over H2 at 2'+ 2" based on direct play is likely to mean something like 30 elo in blitz play against other engines, and 20 in intermediate level play. That would still be very good, but this is about what we did with Komodo in half the time from K4 to K5.
LOL. One of the things the Komodo team members really should learn, is to avoid bad-mouthing and speculation about other engines, in particular about Houdini.
I can disclose that in a direct test against Komodo 5 the current Houdini 3 DEV version scores 62.0%-38.0% (+84 Elo) at 2'+2" TC, 9000 games.
Not sure how this will translate to intermediate levels, we'll have to wait and see...
Robert
Yeah,right.....
And when did you manage to play 9000 games even at this time control when Komodo 5 was released several days ago
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
57.2 implies about 50 ELO which is a lot more than the previous version did so I have to say I am skeptical, but seeing is believing so we will have to wait to see.
Why are you skeptical ?
Well, one reason might be that Robert does not have much credibility, because he refuses to acknowledge that Houdini was (is?) 99% copied code from Ippolit. He could still acknowledge this fact and continue to sell his program, but he doesn't, so why should anyone believe him? Anyway, based on my experience with Rybka and Komodo, a gain of 50 elo over H2 at 2'+ 2" based on direct play is likely to mean something like 30 elo in blitz play against other engines, and 20 in intermediate level play. That would still be very good, but this is about what we did with Komodo in half the time from K4 to K5.
LOL. One of the things the Komodo team members really should learn, is to avoid bad-mouthing and speculation about other engines, in particular about Houdini.
I can disclose that in a direct test against Komodo 5 the current Houdini 3 DEV version scores 62.0%-38.0% (+84 Elo) at 2'+2" TC, 9000 games.
Not sure how this will translate to intermediate levels, we'll have to wait and see...
Robert
Yeah,right.....
And when did you manage to play 9000 games even at this time control when Komodo 5 was released several days ago
9000*8 minutes per game mean 1200 hours of computer time of single processor.
Komodo is available for more than 120 hours
so if you use 2 machines for testing houdini against Komodo(when one of them does 4 games at the same time and another one does 6 games at the same time) then it is clearly possible.
Uri
Last edited by Uri Blass on Wed Jul 25, 2012 10:49 am, edited 1 time in total.
57.2 implies about 50 ELO which is a lot more than the previous version did so I have to say I am skeptical, but seeing is believing so we will have to wait to see.
Why are you skeptical ?
Well, one reason might be that Robert does not have much credibility, because he refuses to acknowledge that Houdini was (is?) 99% copied code from Ippolit. He could still acknowledge this fact and continue to sell his program, but he doesn't, so why should anyone believe him? Anyway, based on my experience with Rybka and Komodo, a gain of 50 elo over H2 at 2'+ 2" based on direct play is likely to mean something like 30 elo in blitz play against other engines, and 20 in intermediate level play. That would still be very good, but this is about what we did with Komodo in half the time from K4 to K5.
LOL. One of the things the Komodo team members really should learn, is to avoid bad-mouthing and speculation about other engines, in particular about Houdini.
I can disclose that in a direct test against Komodo 5 the current Houdini 3 DEV version scores 62.0%-38.0% (+84 Elo) at 2'+2" TC, 9000 games.
Not sure how this will translate to intermediate levels, we'll have to wait and see...
Robert
Yeah,right.....
And when did you manage to play 9000 games even at this time control when Komodo 5 was released several days ago
9000 games at 5 minutes per game is 31 CPU-core days, if you have a few quad-core PCs you can easily do that in two-three days.
rbarreira wrote:9000 games at 5 minutes per game is 31 CPU-core days, if you have a few quad-core PCs you can easily do that in two-three days.
2'+2" translates to about 8 minutes per game.
The 9,000 games run takes less than 48 hours.
Hi Robert,
you said in facebook that H3 could be released in september-october depending by the telescope and the wheather...can you state seriously if you plan to release it really in september or there could be a delay?
you said in facebook that H3 could be released in september-october depending by the telescope and the wheather...can you state seriously if you plan to release it really in september or there could be a delay?
Thank you
Best Regards
Everything is still geared for a September/October release.
you said in facebook that H3 could be released in september-october depending by the telescope and the wheather...can you state seriously if you plan to release it really in september or there could be a delay?
Thank you
Best Regards
Everything is still geared for a September/October release.
Robert
It is surprising a bit, i remember once you let understand that the work in progress of H3 was so good that it was even possible a release in advance (that was in a thread here some months ago), then you stated many times of a release in september and now there's a chance of a release in october.
No problem with it but it is a bit strange in my eyes. Perhaps some problems in the development?
57.2 implies about 50 ELO which is a lot more than the previous version did so I have to say I am skeptical, but seeing is believing so we will have to wait to see.
Why are you skeptical ?
Well, one reason might be that Robert does not have much credibility, because he refuses to acknowledge that Houdini was (is?) 99% copied code from Ippolit. He could still acknowledge this fact and continue to sell his program, but he doesn't, so why should anyone believe him? Anyway, based on my experience with Rybka and Komodo, a gain of 50 elo over H2 at 2'+ 2" based on direct play is likely to mean something like 30 elo in blitz play against other engines, and 20 in intermediate level play. That would still be very good, but this is about what we did with Komodo in half the time from K4 to K5.
LOL. One of the things the Komodo team members really should learn, is to avoid bad-mouthing and speculation about other engines, in particular about Houdini.
I can disclose that in a direct test against Komodo 5 the current Houdini 3 DEV version scores 62.0%-38.0% (+84 Elo) at 2'+2" TC, 9000 games.
Not sure how this will translate to intermediate levels, we'll have to wait and see...
Robert
Yeah,right.....
And when did you manage to play 9000 games even at this time control when Komodo 5 was released several days ago
.