Deep HIARCS 12 v Rybka 2.3.2a UCI 51-49!

[Yarget]

Very impressive performance by the latest Hiarcs version. Check here for more results at the Hiarcs website:

http://www.hiarcs.com/hiarcs_games.htm

Perhaps Hiarcs 12 is a big step forward and close to Rybka regarding playingstrength? I am looking forward to see some results from CEGT and CCRL.

Regards
Per

[Ovyron]

CCRL is going to test HIARCS 12 with Sharpen PV set to OFF, and that will probably underrate HIARCS 12. I find it amusing that they test several Chessmaster settings but that they won't test HIARCS 12 with the recommended settings.

[ThatsIt]

Hi Per !

Very impressive performance by the latest Hiarcs version. Check here for more results at the Hiarcs website:
http://www.hiarcs.com/hiarcs_games.htm
Perhaps Hiarcs 12 is a big step forward and close to Rybka regarding playingstrength? I am looking forward to see some results from CEGT and CCRL.
Regards
Per

Very impressive, indeed.
First results for the CEGT-Blitz-Ratinglist can be seen here:
http://husvankempen.de/nunn/phpBB2/viewtopic.php?t=976
But notice, its only blitz !

Best,
G.S.

[Yarget]

Thanks for the link Gerhard. I'm looking forward to your results as well.

Best regards
Per

[Spock]

CCRL is going to test HIARCS 12 with Sharpen PV set to OFF, and that will probably underrate HIARCS 12. I find it amusing that they test several Chessmaster settings but that they won't test HIARCS 12 with the recommended settings.

I'm glad you find it amusing. The recommended setting was totally unclear. I do not find unclear advice amusing at all, when valuable CPU time is involved. Was it one minute per move on 8 Intel CPUs or an Athlon 1200 ? Huge difference. We only just received an Email from Mark to say he recommends we use ON. We will test BOTH, but since we started with the default of OFF, for my part that is how I will continue. We will do both, that is certain.

[Ovyron]

We will test BOTH, but since we started with the default of OFF, for my part that is how I will continue. We will do both, that is certain.

I'm glad to read that, I'm very interested on how HIARCS will perform with ON and with OFF, and to see if there will be a huge difference, or if it will be insignificant.

[Titu]

Very impressive performance by the latest Hiarcs version. Check here for more results at the Hiarcs website:

http://www.hiarcs.com/hiarcs_games.htm

Perhaps Hiarcs 12 is a big step forward and close to Rybka regarding playingstrength? I am looking forward to see some results from CEGT and CCRL.

Regards
Per

IMHO only 3rd parties like CEGT and CCRL able to do the tests in an unbiased fashion. I have heard a lot talk about "anti-Rybka books" on various places.

/Titu

[Titu]

CCRL is going to test HIARCS 12 with Sharpen PV set to OFF, and that will probably underrate HIARCS 12. I find it amusing that they test several Chessmaster settings but that they won't test HIARCS 12 with the recommended settings.

I'm glad you find it amusing. The recommended setting was totally unclear. I do not find unclear advice amusing at all, when valuable CPU time is involved. Was it one minute per move on 8 Intel CPUs or an Athlon 1200 ? Huge difference. We only just received an Email from Mark to say he recommends we use ON. We will test BOTH, but since we started with the default of OFF, for my part that is how I will continue. We will do both, that is certain.

Totally agree with you Ray. You would expect the default values to be the best to analyse and play games with. That recommendation is confusing from the points you raised. Looks like that the Hiarcs team is focusing too much on fast engine vs engine games(ratings).

[Spock]

Totally agree with you Ray. You would expect the default values to be the best to analyse and play games with. That recommendation is confusing from the points you raised. Looks like that the Hiarcs team is focusing too much on fast engine vs engine games(ratings).

Thanks. We test at default settings, apart from turning off own books and learning. Normally this isn't an issue, but I appreciate that some settings may be better for blitz and some for long time controls. We have a situation here where the default is optimised for blitz, and we need to change something for long time controls as there is a significant performance penalty not to do so. I wonder why the engine could not adapt itself ? When the GUI sends the time control to the engine at the start of a match, it could surely choose the appropriate internal code path ? Anyway, I just wish this had been pointed out far more clearly. We will need to test both at 40/40, or drop the OFF tests if the early results are looking bad - which they probably are

So we have presumably:

CEGT and CCRL Blitz - OFF
CEGT 40/20 - OFF
CCRL 40/40 - ON recommended
CEGT 40/120 - ON

The early CEGT 40/20 people are not affected, very lucky for them

[Spock]

After some further correspondence with Mark, I am going to switch to testing with the parameter set to ON, as it is clear to me now that that should be the priority for me. Then I'll go back to OFF and resume those games