HSHMAX too low when using 3-6 man folder

[hoacin]

Hi,

I'm new on github and not sure this is the right place to post my question. Would be nice if anyone could help me.

I would like to use fathom.exe tool to analyse root positions. I have one issue and one question.

Issue: When I ask for FEN and give 3-5 man folder, program returns pgn as I would expect. When I do the same and give 3-6 man folder, instead of pgn data I get string "HSHMAX too low!". What can I do about that?

I also never found information if winning moves are sorted by dtz. Are they? Can this tool be used to generate guaranteed winning sequence? Playing random winning move would make the winning process infinite.

Thank you and have a nice day.

[basil00]

HSHMAX too low!

This should never happen. Do you have a FEN that exhibits this behavior so I can test? Also which version did you use?

I also never found information if winning moves are sorted by dtz. Are they?

The results are not sorted. It is also difficult to generate "guaranteed winning sequences" from DTZ tables. If you always play the DTZ-optimal move in a winning position, then it is OK. Otherwise you need some way to avoid draw by repetitions.

[hoacin]

Thank you for answer. Here are data that could be usefull to trace the issue:

• Just a weak ago I downloaded here Fathom-1.0.zip and used the windows exe tool.
• 3k4/4n3/8/8/4K1B1/8/1B6/8 w - - 0 1
• My syzygy6 folder has 1020 files and total 161209574528 byes.
• The program returns correct pgn using syzygy5 folder while fails on syzygy6 folder, which has the same files from syzygy5 and the position shouldn't need any 6man tablebases at all.

I can handle losing and drawing (+tablebase) side with UCI, but fastest move to next conversion for winner would certainly help me. Reasoning behind winning process in many endgames is beyond engine horizont. No way to get to this with fathom? Is it hard to calculate or nobody is interested in such a functionality? Why nobody cares about winning a won position with syzygy, when syzygy format is capable of doing that? Chess engines also returns unsorted wins on root. Looking for this functionality already for quite a while, some help with that would be greately appreciated.

[basil00]

Just a weak ago I downloaded here Fathom-1.0.zip and used the windows exe tool.

Okay that's the problem. It seems the github release is very out-of-date, and the HSHMAX bug has already been fixed. I will need to build new versions of Fathom and upload them in the next day or two.