No Masterpiece. The nalimov endgame code is still broken if some sub endgames are missing. Just look at this game, black (houdini 2.0c) had a totally won position but then - due to the buggy code - did not promote. Probably black only had the KPPPKP and not some/all the endgame egtbs for the possible promotions. Rybka handles this perfectly.
Thank you for signaling the issue, but for the moment I cannot reproduce the problem.
The key position is after white move 169.
[d] 8/8/5K2/7p/7P/1k6/pp6/8 b - -
I have loaded all the 5-men Nalimov and the 3+3 6-men tables. So I'm certain to have none of the tables that result after the pawn promotion (as they are 4+2).
Here's the analysis:
No problem here.
Can you show the analysis on your system, and tell me which of the EGTB files with possible relevance for this position are available?
Houdini wrote:Thank you for signaling the issue, but for the moment I cannot reproduce the problem.
The key position is after white move 169.
[d] 8/8/5K2/7p/7P/1k6/pp6/8 b - -
I have loaded all the 5-men Nalimov and the 3+3 6-men tables. So I'm certain to have none of the tables that result after the pawn promotion (as they are 4+2).
Here's the analysis:
No problem here.
Can you show the analysis on your system, and tell me which of the EGTB files with possible relevance for this position are available?
Cheers,
Robert
This is likely a problem where you have the kppkx tables, but not the kqpkx table. There you will see the mate with the pawn on the 7th, but probably not with the new queen which takes you out of tables. There's a known algorithmic fix for this that I developed years ago for this very reason...
3+3 is not relevant here as you need kpppkp. As Bob said: if then kppqkp and all other winning promotions are missing houdini 2.0c fails exactly for the reasons he gave.
IQ wrote:3+3 is not relevant here as you need kpppkp. As Bob said: if then kppqkp and all other winning promotions are missing houdini 2.0c fails exactly for the reasons he gave.
I need more specific data, I don't want to have to guess what is the problem.
Can you please show the analysis you obtain, and tell me which EGTB files you have or have not installed.
Do you have the kpppkptable installed?
If you remove the kpppkp table, does the problem disappear?
A friend of my best friend has Houdini 2.0c Pro !!!!! 74% with Deep Rybka 4.1 in a long match (300 games) ! Never seen before !
I don't know if Jury is fast enough to implement the news in Strelka-Belka 2 MP ! Otherwise Houdini will rest without a brother/sister in top !
Bravo & congratulations ! Maybe His Majesty Dracula changed his mind !
IQ wrote:3+3 is not relevant here as you need kpppkp. As Bob said: if then kppqkp and all other winning promotions are missing houdini 2.0c fails exactly for the reasons he gave.
I need more specific data, I don't want to have to guess what is the problem.
Can you please show the analysis you obtain, and tell me which EGTB files you have or have not installed.
Do you have the kpppkptable installed?
If you remove the kpppkp table, does the problem disappear?
Thank you.
Robert
Yes, removing kpppkp fixes the problem. Also, instead adding kqppkp (in addition to having kpppkp) fixes the problem. No other kxppkp installed. I couldn't find out more, as it is not my machine.
IQ wrote:Yes, removing kpppkp fixes the problem. Also, instead adding kqppkp (in addition to having kpppkp) fixes the problem. No other kxppkp installed. I couldn't find out more, as it is not my machine.
OK, that defines the problem very well.
Thank you for taking the time to report this, much appreciated!