HAL9000

HAL9000
"It just isn't conceivable that you can design a program strong enough to beat players like me."

February 29, 2016

Android UCI engine update: Rodent II JA

Rodent of Pawel Koziol has just gone from v1.7 to v2.0, as promised by the author during the release of Mini Rodent, a temporary but completely clean rewrite which was aiming to prepare a better basis for further progress toward Rodent II.

Previous two versions were rated in Rapidroid as ~2650 ELO for Rodent 1.7 and ~2625 for Mini Rodent, exactly with an expected narrow gap in favor of Rodent 1.7.

Now it's done and it's expected that Rodent II brings considerable improvements. Thanks must go to Jim Ablett for bringing an Android port of it.

My plan is to introduce Rodent II and delete both elder brothers as from April-2016 release of Rapidroid.

Download Rodent II for arm7: HERE
Download Rodent II for arm7: HERE

Android XB engine update: Tucano 6.00.JA

Tucano is not the best of the Winboard/Xboard engines compiled for Android, especially when compared to ExChess, Scorpio and Crafty but it's one of the significant engines.

The updated version 6.00, compiled for Android by Jim Ablett, is expected to replace 5.00 which is currently rated ~2650 ELO in Rapidroid and add some more ELOs.

As i'm just finalizing a new list for March-2015, the new Tucano will have to wait for April to show up.

Let me remind that v5.00 can only work with Chess for Android and only with fixed time per move setting. I don't know yet whether v6.00 will handle Fischer clock.

Download Tucano 6.00 for arm5: HERE
Download Tucano 6.00 for arm7: HERE

Android UCI engine update: Galjoen 0.32.JA

The new version 0.32 of Werner Taelemans' Galjoen is compiled by Jim Ablett for Android.

I must remind here that i could not make the previous Android builds work the way i've been expecting because of irregular cpu usage which seemed uncontrollable by the gui. Technically said: they all didn't stop working on opponent's time and hogging one core.

I hope the new version corrects this issue. For those interested in testing another average engine, the download link is: HERE

Worked hard enough...Back to 64 squares

It took longer than expected for me to get back to what i enjoy from what i must accomplish to ensure the conditions necessary for spending more time doing what i enjoy.

I read again what i just wrote above and it's hard to understand even for the author of the sentence...

Let me try to express in a shorter way: I was too busy at work!

Now that i can (or i believe i can) breathe a little bit and keep on sharing new stuff, i must first confess that it's extremely hard to find a good balance between the possibilities offered by a loaded professional life and the precious amount of time it steals from the remaining part that you always wanna keep enjoyable, fruitful, as pleasant as possible.

I read once again what i just wrote above. If once again, it's hard to understand, nevermind!

I'd better keep posting things one could understand better.

February 13, 2016

SSDF rating list: Update of Feb,10th

SSDF is the oldest of all computer rating lists, featuring a wide variety of platforms, tournament time controls and pondering. Although the hardware they are using is not of the latest technology anymore, the quad core Q6600 is still a reference cpu. At least, it's more powerful than Athlon X2 4600+, the reference CPU of CCRL.

Indeed, the methods SSDF uses to prepare updates of their list has become slower in time, taking into account the speed at the engines are updated today. 10-20 years ago, it used to happen once or twice a year 20 years ago while today the version updates rain almost mothly, even daily when it comes to speak about Stockfish.

No matter what, i respect SSDF and the chairman Lars Sandin for the continuity of the work. What they do since decades is not easy to organise and sustain.

On February 10th, SSDF list has been updated with Komodo 9.1 ahead of Stockfish 6, the former topper. I hope the next list will feature the latest versions of both.

The current Top-10:


# Name                                     Ratg +   -  Gam  Sco Opp
1 Komodo 9.1 MP x64 2GB Q6600 2,4 GHz      3361 30 -27 947  82% 3102
2 Stockfish 6 MP x64 2GB Q6600 2,4 GHz     3328 32 -29 696  77% 3123
3 Komodo 7.0 MP x64 2GB Q6600 2,4 GHz      3269 33 -30 610  74% 3088
4 Komodo 5.1 MP x64 2GB Q6600 2,4 GHz      3241 24 -23 916  65% 3132
5 Stockfish 3 MP x64 2GB Q6600 2,4 GHz     3206 22 -21 1130 66% 3090
6 Deep Rybka 4 x64 2GB Q6600 2,4 GHz       3204 22 -21 1168 70% 3058
7 Deep Hiarcs 14 2GB Q6600 2,4 GHz         3194 22 -21 1138 66% 3080
8 Deep Rybka 3 x64 2GB Q6600 2,4 GHz       3192 22 -21 1371 75% 3001
9 Naum 4.2 MP x64 2GB Q6600 2,4 GHz        3145 21 -21 1083 61% 3069
10 Deep Junior Yokoh x64 2GB Q6600 2,4 GHz 3126 27 -27 650  49% 3126


You can view the complete SSDF list in their homepage: HERE

February 7, 2016

Sting 6 really not good for playing games?

If you often intend to resist against the bare truth until the moment you face the same result after your own practice, you may be one of the victims of that disorder which usually costs nothing serious but your time and sometimes your money. But who cares?

As time and money is available without limit in the time space we are living, this is not a problem at all and let's keep on being sceptical.

Last week i've been hit by another attack of above mentioned disease and i wanted to disprove that Sting 6 is designed for analysis only. The impossible mission dictated that 6.0 should definitely play stronger than 4.84 which is currently rated 3100+ ELO in Rapidroid.

Even if the author says so, he's just the author and he can't know the facts better than an addicted tester. He just can't...

And a long practice started two weeks ago on one of my RK3188 tablets. Even though there was still a chance to prove that i was right and the author was wrong, i had to stop it a few days ago because the latest version (nasty uncooperative thing!) didn't really want to support me on my hypothesis.

First stage with all versions:



1) Sting SF 4.8.4 : 208 / 300
2) Sting SF6      : 198 / 300
3) Sting SF 4.7   : 194 / 300

4) Sting SF 5     :   0 / 300

It's obvious that Sting 5 (incorrectly reports itself as SF4.7) can't play Fischer clock. It exits all the games.

Second stage with 4.8.4 vs 6:

Games 101 to 200 (TCEC7-051 to 100):

Games 201 to 300 (TCEC7-101 to 150):
Games 301 to 400 (TCEC7-151 to 200):
Games 401 to 500 (TCEC7-201 to 250):

Games 501 to 612 (TCEC7-251 to 306):
















OVERALL SCORE:
Sting SF 4.8.4: 342.5 / 612
Sting SF 6.0  : 269.5 / 612


VERDICT: Sting 4.8.4 stays in Rapidroid and v6 goes to the archives. But hey, if i could play 90,000 more games, i could disprove that. Really...

1112 unnecessary games: HERE

February 6, 2016

Toga Returns: Too many fruits on the table...

At first sight, i was happy to receive yet another engine running under Android when Toga Returns had been made available by Jim Ablett.

In fact, there's no reason to oppose a growing rating list as long as enough resources are available for playing test games vs other engines.

That said, i've tested Toga Returns against its cousins. The engine proved to be a strong one, running multi threads and compatible with both time controls used in Rapidroid. It's a flawless port to Android.

Regarding the performance, below summary does not reveal any surprise. Toga 3.0 stands still with a single core while other derivatives Grapefruit, Cyclone and Toga Returns benefit from parallel search rather than better code. Actually, Toga Returns is just a tweaked Toga performing very similar to Grapefruit.
Toga "Returns" 2869 ELO overall but can't crush Grapefruit

The closed tournament was played at 300+1 time controls on  Rockchip 3188 quad core downclocked to 1.0 Ghz. I've used the last 50 openings of TCEC-7. PGN games: HERE

Meanwhile, the number of Fruit derivatives wandering in Rapidroid seemed a little bit high to me and some kind of restlessness forced me jump into dusty Fruit files once again, in order to remember what was what and who forked from what. It's always complicated to understand quickly but i think what i could put together represents the overall forking tree which had started from Fruit 2.1.
Given that i may not be a qualified investigator, in case you notice something incorrect in above structure, let me know about it.

Therefore, the action which finalizes the case must be called "Rapidroid defruitized":
1) Toga Returns has nothing special to add but rating distortion. It must be ignored.
2) Grapefruit stays in because it's the strongest fork so far. I'd better keep that one.
3) Cyclone will be excluded because it's very close to Grapefruit and presents no remarkable difference.
4) TheMadPrune will be excluded because it can't play fixed time per move games and it can't use multi threads although its raison d'etre is to benefit from SMP.
5) Fruit Reloaded stays in because it's relatively new and rewritten.
6) Senpai stays. It's alive, multi threaded, above 3000 ELO and not a Fruit at all.