Winboard Forum

Funny little thing about this strong engine....it can not mate with King and rook, against King....*smiling*

Check for your self..!!!

Leif Rosenb

No, he can solve it. Big trouble in blitz, but at standard time controls, the mate will come.

What is more remarkable is that he solves much faster with K+R than with K+Q.

Try these with Arena and EPD analyze:
3k4/8/8/8/8/8/8/R2K4 w - -
3k4/8/8/8/8/8/8/Q2K4 w - -

Dann Corbit wrote:No, he can solve it. Big trouble in blitz, but at standard time controls, the mate will come.

What is more remarkable is that he solves much faster with K+R than with K+Q.

Try these with Arena and EPD analyze:
3k4/8/8/8/8/8/8/R2K4 w - -
3k4/8/8/8/8/8/8/Q2K4 w - -

Hello Dann,

Is Arena's EPD analysis gunction now working without any hanging issues?

I am desirous of using it for extended EPD analysis but it still hangs after a while here.

Can you suggest anything that might help?

Later.

Roger Brown wrote:

Dann Corbit wrote:No, he can solve it. Big trouble in blitz, but at standard time controls, the mate will come.

What is more remarkable is that he solves much faster with K+R than with K+Q.

Try these with Arena and EPD analyze:
3k4/8/8/8/8/8/8/R2K4 w - -
3k4/8/8/8/8/8/8/Q2K4 w - -

Hello Dann,

Is Arena's EPD analysis gunction now working without any hanging issues?

I am desirous of using it for extended EPD analysis but it still hangs after a while here.

Can you suggest anything that might help?

Later.

What engines do you have trouble with?

I find that if I specify too much time (e.g. I give it one hour and it only takes 5 minutes to solve the problem) then Arena will wait for the full time to expire. I do not know if this should be considered as hanging or not.

Other than that, I have not had any problems using Arena 1.1 and many different Winboard and UCI engines.

Good alternatives are Bruce's EPD2WB (and Thomas Mayer has a better version that has several improvements) and also GradualTest (O.G.M. IIRC)

Dann Corbit wrote:I find that if I specify too much time (e.g. I give it one hour and it only takes 5 minutes to solve the problem) then Arena will wait for the full time to expire. I do not know if this should be considered as hanging or not.

That is due to a flaw in the UCI protocol. When the engine gets a "go infinite" it is not allowed to end the analysis until the GUI sends the "stop" command. Not even in a KBK position, it has to continue and wait the whole hour! It would be better if the engine could return "bestmove xxxx" whenever there is no more to compute.
/Peter

Dann Corbit wrote:
What engines do you have trouble with?

I find that if I specify too much time (e.g. I give it one hour and it only takes 5 minutes to solve the problem) then Arena will wait for the full time to expire. I do not know if this should be considered as hanging or not.

Other than that, I have not had any problems using Arena 1.1 and many different Winboard and UCI engines.

Good alternatives are Bruce's EPD2WB (and Thomas Mayer has a better version that has several improvements) and also GradualTest (O.G.M. IIRC)

Hello Dann,

I find that if I stick to the tried and true engines such as Yace and Crafty in Winboard mode I am alright.

Once I give it to Deep Sjeng - a superb engine - in UCI mode it hangs after a number (not fixed but annoyingly near to the end of the epd I am analysing say one thousand positions) of epd lines are examined.

The pgn analysis follows a similar pattern so I will come back to find that Deep Sjeng (for instance) is hung on a simple postion for hours!

I have hada similar experience using other engines. I typically line up a number of engines - and versions thereof - say over ten and I can count on them not reaching the final engine before Arena hangs.

I will try it with only battle hardened engines report my findings.

I analyse at less than 60 seconds per position - easy positions for a computer.

EPD2WB is rock bottom solid but the batch writing can be a wee bit tedious. I am a spoilt Windowz user....waaaaaahhhhh!



I will have a look at Gradual Test.

Thanks.

Later.