Thinker 5.3B Inert = a new era in computer chess !

[Sylwy]

Hi all !

I have-now- over 1.000 games with Thinker 5.3B Inert. Never I was so impressed from Rybka 1.0 beta !

An efficient and attractive play.
A little executable-only 67.000 bytes.
Playing through an (Wb2Uci) adapter (therefore a 20-30 ELO points loss ).

Only two explanations are possible:

1.or Thinker 5.3B Inert is a raptor in KN/s -therefore it reaches a very deep horizon of search ,or
2.this engine uses some new extensions (a dynamical evaluation, new pruning functions........).
Now Thinker 5.3 B Inert becomes a clear no.1 of the free chess engines.
The danger for Rybka comes from Thinker not from Fritz,Naum,Shredder,Hiarcs........

Oscar 2008 ( or Nobel ) in computer chess for Lance Perkins !
Bravo !

Regards,
Silvian

[Jouni]

Sadly it's only WB and don't display any analysis lines. Even if it were at Rybka 3 level, I probably won't download.

Jouni

[Steve Maughan]

Sadly it's only WB and don't display any analysis lines. Even if it were at Rybka 3 level, I probably won't download.

I agree. UCI is the standard - Winboard is yesterday's technology. It's the computer chess equivalent of releasing a record only on vinyl. Look, even the testers are using wb2UCI to test the engine. Come on Lance, give the computer chess community what they want

Cheers,

Steve

[kranium]

My 40/4 (CCRL blitz time control) test matches seem to indicate that your enthusiasm may not be unfounded. Let's see:

25-7 against BugChess2 1.6.3
16-16 (!) against Naum 3.1
26,5-5,5 (!) against Spike 1.2 Turin

Note: This was the MP version running on 2 threads.

As for the small size, I think Thinker is written in machine code.

Hi Gabor-
I'm very interested to see your results, thanks for posting them.
I do have a question:

I'm not sure how to interpret it when an engine using 2 cpus beats an engine using 1?

for ex: why wouldn't one simply test 1 cpu against 1 cpu, or 2 vs 2?

is there a way to draw a conclusion concerning relative strength with these results?

Thanks in advance,
Norm

[kranium]

My 40/4 (CCRL blitz time control) test matches seem to indicate that your enthusiasm may not be unfounded. Let's see:

25-7 against BugChess2 1.6.3
16-16 (!) against Naum 3.1
26,5-5,5 (!) against Spike 1.2 Turin

Note: This was the MP version running on 2 threads.

As for the small size, I think Thinker is written in machine code.

Hi Gabor-
I'm very interested to see your results, thanks for posting them.
I do have a question:
I'm not sure how to interpret it when an engine using 2 cpus beats an engine using 1?
i.e. are test results between engines not using the same # cpus indicative of strength?

for ex: why didn't you just test 1 cpu against 1 cpu, or 2 vs 2?

this is absolutely not criticism...i just don't get it.

Hi Norman,

In my testing I use MP-engines on 2 cores. This is partly to check they run flawlessly on multi cores, partly to reward them for being multi-core capable, and partly because for the CCRL group no one else contributes 2-CPU blitz games.

were Bugchess, Spike, and Naum, also running on 2 cores?
If so, i simply misunderstood the post....

Thanks
Norm

[ArmyBridge]

Hi all !

I have-now- over 1.000 games with Thinker 5.3B Inert. Never I was so impressed from Rybka 1.0 beta !

An efficient and attractive play.
A little executable-only 67.000 bytes.
Playing through an (Wb2Uci) adapter (therefore a 20-30 ELO points loss ).

Only two explanations are possible:

1.or Thinker 5.3B Inert is a raptor in KN/s -therefore it reaches a very deep horizon of search ,or
2.this engine uses some new extensions (a dynamical evaluation, new pruning functions........).
Now Thinker 5.3 B Inert becomes a clear no.1 of the free chess engines.
The danger for Rybka comes from Thinker not from Fritz,Naum,Shredder,Hiarcs........

Oscar 2008 ( or Nobel ) in computer chess for Lance Perkins !
Bravo !

Regards,
Silvian

Yep, total agree. I've been think about Junior11 release day, but no more we have a new King of destruction!!

[S.Taylor]

Sadly it's only WB and don't display any analysis lines. Even if it were at Rybka 3 level, I probably won't download.

Jouni

If it can be shown to be atleast as strong as Rybka 3, AND an attacking style, then I'd consider it as even stronger than Rybka (because it is even more open to challenge, if it is more aggressive too), and, I'd not be without it, even if above average price.

I'm sure that minimal analysis lines can be seen, but if its strength is so fantastic, it would be a living legend even if not.

However, we DO prefer the updated bells and whistles and compatabilities, as well.

[fern]

In fact, whistles and bells are of the esence these days. Strenght is already a comodity. I will not expend a single cent for a stronger engine than the many very strong I have now. For what?
Perhaps that has sense for people that use engines to analyze his games because they play in tournaments, but I just play programs.
Even a good dedicatet unit suffices to me.

My best
fern

[BubbaTough]

Anyone know which version of of Thinker participated in the ACCA tournament? Very respectable results, only losses were to Rybka and Naum (with a draw with LearningLemming, but that program is practically unbeatable anyway). It was able to show some PV by the way, which I believe was extracted out of the hash table. While this is not the same as a "Real PV", I would encourage the author to consider including this feature in normal releases since extracting PV from the hash table imposes no overhead in the search, and produces a PV good enough to please most users.

-Sam

[Ovyron]

If it can be shown to be atleast as strong as Rybka 3, AND an attacking style, then I'd consider it as even stronger than Rybka (because it is even more open to challenge, if it is more aggressive too), and, I'd not be without it, even if above average price.

For the record, Rybka 2.3.2a used to play some boring and dull chess, but Rybka 3 no longer does, I could argue that Rybka 3 Human's playing style is aggressive.