HAL9000

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

March 29, 2015

New Android UCI engine: Kurt 0.9.2.2 beta JA

Don't be fooled by beta state. Kurt is not a recent engine. It was written by Oliver Uwira from Germany relatively long time ago.
The version ported to Android by Jim Ablett just today was originally released in 2010. Regarding the strength, CCRL rates v0.9.2.2 at 2173 ELO on a PC. I guess the beta version on Android will play around 1900.

Download Kurt: HERE

March 26, 2015

30000

The day i started to blah blah about comp chess here, i could not expect such specific place would attract some attention. In less than one year, it already makes:
30,000 hits !!!
I never pushed forward anything to boost visitor hits and there have been only a limited number of posts here and then in some chess forums, a few posts sent months ago to Facebook (which i'd unregistered later on) and that's all.

Today, only Googlers fall into Chesstroid. I'm not against. It's okay. That's what makes me happy because i'm not interested anything other than sharing things i'm doing with chess addicted people willing to see them.

I really dislike everything that the commercial trends try to inject into our eyes and brains to make more money and get full control of our pockets. Do you?

Android XB engine update: Purple Haze 2.1.0 JA

A new build of Purple Haze by Jim Ablett promises better compatibility under Chess for Android. There used to be a very old build of this version by JA made in November 2012. I was much bigger in size above 1MB. Unfortunately i never managed to have it running fine, thus i couldn't push it into the Rapidroid ranking. I hope the new build will run okay.

The engine is written by Vincent Olivier. It's not a strong one with only 1873 ELO on a PC according to CCRL 40/4 list but Haze makes part of a wide n' colorful palette that i'm addicted to test and rate.

Download Purple Haze: HERE

PS: First tests reveal there are still time management issues. PH got flagged in all Fischer clock games. It uses more and more time on each move. Same applies to fixed time per move. Not eligible for tourneys.

Android UCI/XB engine update: Sjaak II 1.0 JA

Sjaak II finally came to 1.0 after many betas. We used to have beta5 compiled by Jim Ablett. Now, Androiders can use the same latest version as Windows.

I've tried to import Sjaak to Chess for Android using UCI. It loaded fine but i need test games to confirm full compatibility for tournaments later on. Glad to see Sjaak providing UCI too.

In any case, and by default, whenever we have an engine which claims UCI and Xboard/Winboard compatibility, we must absolutely try UCI first.

One main strict reason is: Hash size, number of threads and tablebases can only be configured with UCI protocol but only default settings can be used with XB. This means that on Android, all XB engines lose more strength than UCI. Take UCI when available as a rule.

Also, i must mention that the beautiful alternative GUI Droidfish can NOT use XB engines at all.

Download Sjaak: HERE

March 25, 2015

New Android UCI engine: Maverick 0.60 JA

Hello again to new chess engines after a few days of silence! Maverick is the follow up of the parade.

Maverick is an engine under active development by Steve Maughan. It's open source, meaning that Jim Ablett didn't miss the chance to compile it for Android.

I didn't test it yet as all my devices are fully dedicated to Rapidroid experiment. It uses UCI protocol and plays around 2360 ELO according to its author. I hope it becomes a rising monster soon and join the toppers. Regarding the Android version, it should perform around 2200 roughly.

Download Maverick: HERE

March 20, 2015

Strongest Android chess engine?

It's been one year since i'd submitted a post named like the above. I couldn't foresee it would remain the most displayed post ever in this blog. That's weird but true. Nothing matters but keywords, no matter whether they were used on purpose or not.

To me it's shocking to see people still googleing for something obvious. If you wanna know the big secret, let me tell that the original compile of Stockfish 6, available on stockfishchess.org, is the strongest Android engine to date that one may run in his device. Dot. Finito.
You don't believe that? Ok. First you may follow Rapidroid results here in this blog to be sure. Otherwise, you can launch a thousand game match between Komodo 8 and Stockfish to find the same. Your own choice, your processor, your time.

I wanted to come back to the same subject again under the same post title, because i don't want people to jump into a dusty n' rusty topic sent one year ago which states an older version. Reading this one is much better and up to date. Meanwhile, you'd better stop googleing for the strongest engine unless long awaited Komodo 9 comes up.

Low budget hardware comes help Rapidroid go faster


At last i've found what i've been looking for. Although octacore processors on 64 bit Android are  attractive for computer chess, every rose has its thorn and you must pay the price if you wanna carry a supersonic pocket chess monster... Again these awful technometrics! Well is it really worth the fire in the wallet?

Let's suppose one doesn't want to obey commercial equations of the market and still insists in grabbing more computing power, at least above average without paying more Typically, that's me.

If the idea is to boost Rapidroid for example, what i need is a decent processor with minimum 32K level-1 cache and 2GB RAM and nothing elde in fact. Anything other than the processor and the memory is irrelevant for chess. I needed a tablet without wireless connections, without huge HD IPS screen, no ultra-megapixel camera.

Unfortunately such custom stripped spec with a good cpu is a rare combination. I had to dig around many online stores before deciding to buy that Chinese tablet branded Codegen. It has a quad core Rockchip 3188 processor running at 1.4Ghz (and not 1.6 as seen on the ads!), 2GB RAM, 16GB storage.

No bluetooth, just wifi, a very bad TFT screen with 1024*600 resolution from mid age, plastic rubbish case, unstable tiny buttons... All for around 100 USD. I still think it's the best price/perfo selection for something supposed to do nothing but play chess continously. Besides, it comes with the standard micro USB charge instead of an ugly charger. So i can charge it with my external battery too.

Following the online purchase, i've got it rooted after a few tries, installed Chess for Android, Droidfish. Then, added my specific monitoring utilities and launched the first set of round robins already. As expected, Rockwell performs very similar to Exynos 4412 except for the frequency. It's sad Rockwell silently underclocked 3188 to 1.4GHz. However, it's very proportional anyways and  it runs as stable as its rival. KNps values show 14 to 16 ratios with almost all engines i've tested.

I've also played the same gauntlet of 20 games on 3188 and Exynos 4412. They issued almost identical results. It's fine as a new device for Rapidroid.

I'm glad it's still possible to pay less for more in this commercial era but let me repeat it's a personal preference. One can always choose to invest 10 times more and just double the nodes per second. Just a matter of taste.

March 19, 2015

Android UCI engine update: Rodent 1.7 JA

This may be the quickest port of an engine to Android platform ever because the update of Rodent has been made publicly available less than one day ago.

The previous version 1.6 was also compiled by JA and used to play at ~2600 on Android. We can expect nothing but some more from this latest 1.7 version.

Download Rodent: HERE

New Android XB engine: Carnivor 1.0 JA

In addition to Romichess which was already ported to Android recently, Carnivor is another engine by Michael Sherwin. I have no idea how strong it is but probably it's not a topper. Some sites rate it around 1500 on a PC. Therefore i leave it to the attention of other testers before i give a try.

Download Jim Ablett's compile: HERE

March 18, 2015

New Android XB engine: Evolution Chess 0.6 JA

Enuff silence. Here we go with more and more Android compiles by Jim Ablett. Looks like he's done with all above average engines with available source code so that he's started to dig deeper for more to find. Note that it starts to get difficult for testers because these typical not-yet-mature engines are often subject to early bugs. Caution...

Evolchess is written by Bhupendra Aole. Evol is not rated yet by CCRL but i've found some other sources stating it around 1200 ELO. Anyways, i can't presume Evolchess is necessarily one of the headache engines before conducting any tests. I hope it works perfect under Android and i'm more than willing to deal with another weak engine. Here, no problem at all.

You may download Evolchess: HERE

March 14, 2015

SlowChess 2.96e and Octochess r4995 excluded from Rapidroid

Bullsheet! SlowChess fails in Rapidroid games in contrast to previous flawless test games. Unfortunately, i have no workaround for such bad terminations. Octochess r6556 will come to replace it.

Similar kind of bad terminations apply to Octochess r4995 too. They were not illegal move errors but unexpected exits i can't explain with dead draws or obvious wins. If so, it's possible to adjudicate like what rarely happens with Black Mamba for example.

Let's relate it with SMP implementation of this one and switch to r6556 without SMP. By the way, i must admit r4995 was way stronger with SMP vs r6556 in normally ended games i've played. If only it was stable...

Here's one of the games which caused exclusion of SlowChess in Rapidroid:

[Event "Rapidroid"]
[Site "R01D05"]
[Date "2015.03.10"]
[Round "59"]
[White "SlowChess 2.960e"]
[Black "GarboChess 3 (32-bit)"]
[Result "0-1"]
[PlyCount "82"]
[TimeControl "600+2"]
[Device "GT-N7100"]

1. d4 Nf6
2. c4 e6
3. Nf3 d5
4. Nc3 c6
5. Bg5 h6
6. Bxf6 Qxf6
7. e3 Nd7
8. Bd3 dxc4
9. Bxc4 g6
10. a4 Be7 {from PGN 0.006}
11. a5 {[10]52/e8g8 17.411} O-O {[12]-24 20.09}
12. O-O {[10]52/f8d8 17.38} Rd8 {[12]-26 19.404}
13. Ne4 {[10]44/f6g7 13.094} Qg7 {[13]-37 18.856}
14. Qc2 {[10]36/c6c5 13.184} Rb8 {[12]-26 18.243}
15. Rfd1 {[9]60/b7b6 13.278} Nf6 {[11]-32 17.731}
16. Nc5 {[9]60/f6g4 13.265} Nd5 {[11]-34 17.231}
17. e4 {[10]56/d5f4 13.37} Nf4 {[12]-20 16.748}
18. Nd3 {[9]56/e7c5 13.324} Nxd3 {[11]-16 16.228}
19. Rxd3 {[9]52/c8d7 3.608} Bd6 {[10]-4 15.773}
20. Bb3 {[9]56/d6f4 12.078} g5 {[10]-8 15.326}
21. Nd2 {[9]64/d6b4 12.791} Qg6 {[10]10 14.836}
22. f3 {[10]64/d6f4 13.967} Be7 {[10]0 14.392}
23. Nc4 {[10]56/g6g7 13.929} Qg7 {[11]-25 14.053}
24. Qf2 {[10]52/e7f6 14.043} Bb4 {[11]-25 13.597}
25. Ne5 {[10]64/b4d6 14.159} f6 {[11]-9 13.255}
26. Ng4 {[10]68/h6h5 12.387} h5 {[10]6 12.883}
27. Ne3 {[10]72/g8h7 12.415} Kh8 {[10]2 12.473}
28. Qg3 {[9]96/c8d7 14.412} e5 {[11]-13 12.144}
29. Nc2 {[10]64/b4d6 14.444} Bd6 {[12]24 11.847}
30. Qf2 {[11]52/e5d4 14.548} exd4 {[13]62 11.489}
31. Rxd4 {[9]52/h8h7 6.56} Qc7 {[12]63 11.203}
32. g3 {[9]28/h8h7 6.628} c5 {[13]71 10.852}
33. Rda4 {[9]36/c8d7 7.262} Bd7 {[12]109 10.597}
34. Rc4 {[10]-16/b7b5 6.74} b5 {[12]86 10.315}
35. Rc3 {[10]-32/c5c4 6.67} c4 {[11]135 10.076}
36. Ne3 {[10]-60/d7c6 8.873} Bc6 {[11]148 9.736}
37. Bd1 {[10]-68/d6b4 6.718} Bc5 {[10]156 9.467}
38. Kh1 {[8]-60/c5b4 6.854} Qd7 {[10]182 9.244}
39. Qe2 {[9]-96/d7d2 6.917} Qh3 {[10]200 9.085}
40. a6 {[8]-48/h5h4 6.864} b4 {[11]215 8.842}
41. Rc2 {[9]-65/c5e3 6.486} g4 {[9]238 8.525} 0-1
{white resigned:illegal move b2c3 ponder c5e3 }

March 12, 2015

New Android UCI engine: Smash 1.03 JA

Smash is not a rising star at all. It was last updated in 2006. No matter how old it is, 2015 brings an Android compile by Jim Ablett. We should not expect too much from it but that doesn't mean a new piece in the collection will be rejected. It will soon to be rated in Rapidroid in case further tests validate its compatiblity with engine tournaments.

If you wanna test it before me, go ahead: HERE
The author Maurizio Sambati thinking
(about the next release of Smash?)

March 11, 2015

Rapidroid gets wider, if not deeper!

New and updated engines come falling on our heads everyday. These days are as sad as happy. Weird statement, huh? Indeed, it's fun to see the pool gets bigger but you've got to face the truth that every update needs a rework of the past games and standings.

Every tester has its own methods to deal with updates, Some accumulate new versions over olders and keep old games just the way FIDE treats human players and some others start from scratch after each update, erasing previous versions and games, just the way a gardener deals with the grass generations spilling over and over. You just cut'em and wait for new ones.

A third method is to never remove any previous result and keep all versions together, like does CCRL. That ends with thousands of versions together, a much bigger database and requires special care for avoiding games between versions of the same engine, in order to prevent rating distortion

I'm simply going for the start-over method for Rapidroid, meaning that in time many games will have be thrown away and replaced with new ones. A little bloody, destructive but simplest way.

As of today, below engines have been certified good for tournaments after last Rapidroid release. They currently play gauntlets to find a place in the next ranking. I hope to submit it by the end of next week with about 2000 games total. The main harm with so frequently updated engines is that the number of games played per engine can't increase as quickly as expected. When there's no updates, we can proceed to play more games per engine to shrink the error margins. When updates are raining, it's unavoidable to spend time to refresh everything and delaying next rounds. Wider population means wider margins.

Finally, new entries will be:
* Joanna (weakest engine ever?)
* Mango Paola
* Murka (Unexpected puncher! A good one...)
* NGplay
* Protector (Looks much weaker than on a PC)
* Slowchess
* Trappy Beowulf
* Vajolet
* Colossus C64
* Colossus C64 200x (100x seemed not enough to me lol...)

Updates:
* Sayuri 2015.03.02
* Sting 4.8.4

That means the next release will issue 104 + 10 = 112 engines at least.

New Android UCI engine: Ecce r508.JA

Ecce is an average engine for the moment. It's developed by Denis Gruzdev and Maxim Melnikov from Russia and has a rating of 2002 at CCRL as per 40 moves in 4 minutes time controls.

I would expect Ecce to reach 1850 ELO on an Android device. If you're a helpless collector of engines, don't hesitate to give it a try after downloading: HERE

New Android UCI engine: Clubfoot r2293.JA

A newborn, recent chess engine by Shawn Chidester comes to Android thanks to Jim Ablett. I hope it works well and gets eligible for Rapidroid.

It's interesting to discover that the engine developer has only released the source and not any compile. Perhaps, Clubfoot is the first engine in history which has been compiled for Android before PCs.

There are a dozens of engines still being tested with gauntlets for introduction in the rating list. That's why i won't be able to feedback more info about Clubfoot very soon. JA is so quick to compile on-the-fly that there's always a waiting list on my Android chess todo's.

Meanwhile you may already download it and try before me on your devices: HERE

Money?

They said: "Wow, u got 7000 visits per month to your blog. U should consider making money of it."
I said "Wow, u got 7000 capitalist ideas per hour. U should consider staying away from me."

March 9, 2015

1.d4 by the shore

A cliché i could not keep myself from,
done last month during a weekend trip

Android UCI engine update: Octochess r4995.JA & r6556.JA

The search goes on toward a solid rock Android compile of Octochess. Jim Ablett has just compiled two more alternatives, of which one at least should work flawlessly.

Octo is originally a multi thread capable engine on PC but it seems it had specific flaws under Android so far as previous compiles didn't work on my Arm7 devices, although i could configure the number of cores and even run the engine. Unfortunately, it kept failing during test games.

The latest compiles are r4995 with multicore enabled and r6556 with multicore disabled. Try both and take the one that suits your case.

Android UCI engine update: Ifrit m1.8 JA

I've got a very old compile of Ifrit again by Jim. It never worked for me. I hope the new version, with the help of the polyglot adapter as an XB engine, will perform better.

Same rules as Protector and Pawny apply here as well: You don't rename the ifrit.ini configuration file which is coded into the provided custom Polyglot adapter and you will install, then import the adapter only to the GUI.

I renamed the real engine file as ifrit.eng to prevent confusions when choosing the engine under Chess for Android and modified the engine ini accordingly. I've also specified 128MB hash.

For devices with 1GB RAM, i recommend 256MB hash when using one engine and 128MB for engine tournaments. On devices with 2GB RAM or more, you can simply double above values.

Download Polyglotized Ifrit for Android: HERE

March 8, 2015

Updated summary of all Android engines

Here's the latest list of all Android engines prepared in Excel-97 format, with all specs i could collect. All engines available at this exact moment (!) should be here.

You are welcome to help me get it as accurate as possible with other engine i may have missed or any incorrect data.

Please note that CCRL ratings are added when i first create a new engine entry and remain the same. They are not meant to be updated afterwards.

New Android XB engine: NGplay 9.86.JA

NG is a moderate engine written by George Georgopoulos from Greece, reaching 2180 ELO in CCRL rankings. Thanks to Jim Ablett, we have now NGplay v9.86 for Android Enjoy it!

Android UCI engine update: Sting 4.8.4.JA

This is the latest version of Sting which replaces 4.7 which was previously compiled by Jim Ablett and posted here as well.

Sting is a Stockfish derivative, forked from SF 2.3.1 by Marek Kwiatkowski from Poland. It's regularly updated since the beginning. I guess it will get much stronger in time. Today it has almost 3100 ELO on CCRL scale.

 Android? You can follow Rapidroid for that. Sting 4.8.4 will be very soon called for duty.

Android UCI engine update: Octochess r6555.JA

I'm glad to see Octo getting an update because previous one, as seen in Rapidroid ranking, has something strange, possibly due to a compiling issue, which caused it to play way weaker than expected, despite the multi core setting available and functional.

I hope the new version compiled by Jim helps Octochess get what it deserves. Must hit 2600 ELO at least on Android.

P.S.: Tried UCI and XB both on my RK3066 device today. After import, i see "engine exits"... :-(

New Android XB engine: Mango Paola 1.1.JA

Do you like Mango? I don't. But i would when it's a Mango which plays chess.

Mango Paola was written by Jose Andres Morales Linares from Venezuela. It had no updates since a while and CCRL rates v1.0 with 2236. Quieremos unas nuevas versiones amigo!

We have an Android compile by Jim Ablett which, i expect to be near 2000 ELO. We'll see about that if it passes the tests.

Download Magno Paola: HERE

New Android UCI engine: Vajolet 2.03.1.JA

Vajolet is written by Marco Belli from Italy. It makes almost two years since there's no updates. Could be that the engine is not maintained anymore.

Although not very recent, Vajolet is not a weak engine. On CCRL it has 2658 ELO. On Android it might be up to 2500.

Download Vajolet: HERE

RAPIDROID: Games played so far with full logs

As per a visitor request, i've uploaded all 1556 games played so far in Rapidroid. The file is a 1.4MB 7z archive and contains one PGN with all games and all engine PV logs. I'm not IBM and no reason to hide the logs :-)

I think it can be also useful to check depths reached, terminations etc. and compare across devices. Enjoy!

RAPIDROID GAMES IN PGN

Protector 1.8 & Pawny 1.0: Packages uploaded again

I've just uploaded a third time, the zip packages of two Polyglot supported engines, Pawny and Protector with corrected filenames and 128MB hash. The links don't change, but let me give'em again:

The tests confirm we can not alter the names of ini files. They must remain exactly the same. Even replacing p with P makes the engine unusable.

Hopefully, it's possible to rename the polyglot binary. It's also possible to copy the real engine binaries, as well as the ini files directly to Chess for Android eng folder. I tried to install them via the GUI and copying them directly. In both cases, the result is the same. All 3 files show up in the engine list and we don't have to import the real engines.

Hash and thread settings in ini files work for both Pawny and Protector. Pawny is not multithreaded but Protector is. I've got Protector to reach 506 kNps with 4 cores on Exynos 4412.

I've just finished testing Protector vs Robbolito with 120+1 and 600+2 games. Unfortunately, Protector causes termination problems. It hangs randomly without reason, especially at 52th move (?). Why 52?

I've been checking all terminations where CfA reported "White (Black) terminated". The PV scores show nothing adjudicatable.

I'll keep testing Protector with fixed time per move and Pawny.

March 7, 2015

RAPIDROID RELOADED: March-2015

Here i sail away with the wind blowing! A big second round has almost doubled the number of samples and has pushed the number of rated engines to a huge 104.

I agree 104 engines is nothing compared to CCRL and many others but one can't simply ignore countless tourney management features still missing on Android. I wish i could auto-adjudicate, use random ctg books etc. When i compare the current size to 56 engines of the previous experiment playing at 15 sec/move, the reloaded version of Rapidroid is already much bigger and has potential to challenge its brothers maintained on PC.

Following the first release posted in February, i've been suffering a lot trying to introduce many new engines compiled by Jim Ablett using the modern Fischer time controls per game, specifically 10'+2" to balance speed vs game quality.

It quickly got complex because of 3 typical engine profiles:
* Engines compatible with both fixed sec/move and Fischer clock = The good allies!
* Engines compatible with fixed time per move only = Pairing headache, often longer games
* Engines compatible with Fischer clock only = Pairing headache when above ones are present.

The worst is that different time control capabilities lead to impossible pairings. That means round robin divisions can't issue equal number of games, neither per division, nor per engine and playing a division of let's say 10 engines in a single automated run is not possible anymore either. Gauntlets are necessary to complete all possible pairings.

Anyway, i've concluded it's worth the pain because covering the maximum area remains attractive. Finally, let me stop the non-sense self-torture tears and find below the final standing i've obtained so far.

Don't pity me too much because i'll run 3 devices in parallel as from next round, targeting 6 games/hours average.

BAYESELO BASED ON 1556 GAMES PLAYED BY 104 PROGRAMS
Rnk Name                     O/S T  Elo   +   - ga sco oppo dra
  1 Stockfish 6              A32 4 3153 104  94 36 79% 2952 36%
  2 Komodo 8                 A32 4 3132 104  95 36 75% 2954 33%
  3 Sting SF 4.7             A32 4 3020 101  95 36 68% 2883 31%
  4 Firenzina 2.4.1 xTreme   A32 4 3017  90  88 36 58% 2967 50%
  5 BeeKay                   A32 4 2988  87  87 36 53% 2970 61%
  6 BlackMamba 2.0           A32 4 2981  87  87 36 51% 2971 58%
  7 Texel 1.05               A32 4 2970  91  92 36 50% 2972 39%
  8 Critter 1.6a             A32 4 2949  88  89 36 46% 2974 53%
  9 Cheng4 0.38              A32 4 2923 105  98 34 71% 2783 29%
 10 DeepSaros ver.2.3f       A32 4 2898 101  95 34 66% 2786 32%
 11 Komodo 3 AB              A32 1 2893  91  91 36 50% 2897 44%
 12 RobboLito 0.085e4l       A32 1 2883  97  96 34 50% 2877 35%
 13 Arasan 17.5              A32 4 2796  98  98 34 50% 2798 24%
 14 Senpai 1.0.JA            A32 4 2786  95 100 34 35% 2888 41%
 15 Doch 1.3.4 JA            A32 1 2741  97 101 34 40% 2805 26%
 16 Cyclone 3.4              A32 1 2740  97 104 34 28% 2893 38%
 17 Gaviota v1.0-d           A32 4 2733  97  96 34 50% 2728 35%
 18 Deep Saros 0.9           A32 1 2731 112 109 32 59% 2652 19%
 19 Fruit Reloaded 2.1       A32 1 2730 104 116 36 25% 2915 17%
 20 Grapefruit 1.0           A32 4 2698 100 101 34 46% 2732 21%
 21 Hakkapeliitta dev 120115 A32 1 2662 105 101 32 63% 2578 25%
 22 Toga II 3.0              A32 1 2661  98 102 34 40% 2736 26%
 23 DiscoCheck 5.2.1         A32 1 2655 110 105 32 59% 2579 13%
 24 Scorpio_2.7.7.JA_xb      A32 1 2651 139 116 28 86% 2389 21%
 25 RedQueen 1.1.97          A32 4 2641 106 101 32 64% 2540 28%
 26 Rodent 1.6 (build 6)     A32 1 2592 104 102 32 56% 2546 25%
 27 Deuterium v14.3.34.130   A32 1 2575 100 100 32 48% 2589 28%
 28 IvanHoe 9.46b            A32 4 2569 106 112 32 36% 2672 22%
 29 Crafty_24.1.JA_xb        A32 1 2545 107 102 32 63% 2454 19%
 30 Gull 1.2.JA              A32 1 2530  99 101 32 44% 2554 31%
 31 Rhetoric 1.4.1           A32 1 2527 107 103 32 59% 2456 19%
 32 GNU Chess 5.50           A32 1 2481 101 102 32 47% 2502 25%
 33 TheMadPrune 1.7.04       A32 4 2470 111 113 28 46% 2496 14%
 34 Tucano_5.00.JA_xb        A32 1 2458 106 104 28 55% 2416 39%
 35 Alfil 12.10              A32 1 2394 106 106 32 48% 2397 16%
 36 Daydreamer 1.75 JA       A32 1 2381 100 101 32 48% 2399 28%
 37 Danasah_5.07.JA_xb       A32 1 2372 105 103 28 54% 2361 36%
 38 Bison 15.1               A32 1 2371 103 112 32 30% 2516 22%
 39 Rotor 0.8                A32 1 2352 106 114 32 33% 2478 16%
 40 Sloppy_0.23.JA_xb        A32 1 2351 106 107 28 48% 2364 25%
 41 Dirty_030411.JA_xb       A32 1 2341 103 104 28 46% 2365 36%
 42 Phalanx_XXIV.JA_xb       A32 1 2334 113 109 28 59% 2272 18%
 43 Jazz Orchestra 8.40      A32 4 2330 123 106 36 86% 2027 22%
 44 BBChess 1.3b JA          A32 4 2314 110 126 32 23% 2523  9%
 45 DanasahZ_0.4.JA_xb       A32 1 2298 109 110 26 46% 2322 31%
 46 Pepito v1.59             A32 1 2292 107 112 30 40% 2362 13%
 47 GreKo_12.5.JA_xb         A32 1 2285 108 108 28 50% 2279 21%
 48 GarboChess 3             A32 1 2281 105 105 30 50% 2280 27%
 49 Typhoon_1.0.r358.JA_xb   A32 1 2273 112 109 26 56% 2224 27%
 50 Natwarlal_0.14.JA_xb     A32 1 2257 113 109 28 59% 2190 18%
 51 BetsabeII_1.47.JA_xb     A32 1 2250 110 109 26 54% 2228 31%
 52 Glaurung Mainz           A32 1 2217 136 166 22 20% 2457  5%
 53 Sungorus 1.4 JA          A32 1 2205 103 100 32 59% 2143 25%
 54 Amy_0.8.JA_xb            A32 1 2205 108 108 30 50% 2202 13%
 55 Simplex 0.9.8            A32 1 2194 110 107 34 59% 2104  6%
 56 Rattate_Nosferatu.JA_xb  A32 1 2189 110 109 30 52% 2173 10%
 57 TJchess 1.1U             A32 1 2178  97  97 34 53% 2133 35%
 58 Scidlet_2.61b2.JA_xb     A32 1 2168 116 110 28 61% 2084 14%
 59 Olithink_5.3.2.JA_xb     A32 1 2153 103 103 32 52% 2149 16%
 60 Diablo 0.5.1b JA         A32 1 2146 104 106 30 45% 2179 23%
 61 Myrddin_0.86.JA_xb       A32 1 2144 114 110 28 57% 2087 14%
 62 Resp_0.19.JA_xb          A32 1 2122 103 101 36 57% 2050 19%
 63 CPW-Engine 1.1.11        A32 1 2122 111 104 36 69% 1949 17%
 64 Robocide 281214.JA       A32 1 2059 118 104 36 79% 1828 14%
 65 DanChess_1.04.JA_xb      A32 1 2041 110 122 30 30% 2193  7%
 66 KmtChess_1.21.JA_xb      A32 1 2038 118 118 26 48% 2049 19%
 67 Gk_0.90.JA_xb            A32 1 2033 163 171 16 56% 1918 13%
 68 Romichess_P3K.JA_xb      A32 1 2015 102 111 32 28% 2166 25%
 69 Surprise_4.3.b13.JA_xb   A32 1 2012 176 154 16 69% 1863  0%
 70 Knightcap_3.7F.JA_xb     A32 1 2003 113 116 26 44% 2054 27%
 71 Ges_1.34.JA_xb           A32 1 1999 130 126 22 55% 1973 18%
 72 Samchess_JA_xb           A32 1 1982 106 115 30 30% 2128 20%
 73 Faile_1.44.JA_xb         A32 1 1958 120 115 26 60% 1888 12%
 74 Woodpecker_2.11.JA_xb    A32 1 1956 113 114 26 48% 1973 27%
 75 Fairymax_4.8l.AB_xb      A32 1 1949 133 137 18 42% 2024 28%
 76 ZCT-0.3.2500             A32 1 1939 103 103 36 49% 1960  8%
 77 BikJump v1.8             A32 1 1918  99 101 36 44% 1962 17%
 78 Witz_Alpha21.JA_xb       A32 1 1914 115 110 28 63% 1823 25%
 79 Leonidas_r83.JA_xb       A32 1 1882 114 116 26 46% 1900 15%
 80 Cilian_4.14.JA_xb        A32 1 1867  92  94 36 44% 1906 28%
 81 Colchess_8.0.JA_xb       A32 1 1844 111 111 28 54% 1832 29%
 82 Claudia v. 0.5           A32 1 1830 147 158 16 38% 1909 13%
 83 AdroitChess0.4 JA        A32 1 1824 104 113 36 36% 1920 11%
 84 Umax_48w_xb              A32 1 1801 125 145 22 23% 2009 18%
 85 Sjeng_1.12.JA_xb         A32 1 1796 126 124 24 52% 1783  4%
 86 SjaakII_beta5.JA_xb      A32 1 1796 120 119 24 52% 1783 21%
 87 Bismark_1.1              A32 1 1780 101  96 36 63% 1689 31%
 88 Zzzzzz_3.5.1.JA_xb       A32 1 1663 111 116 26 38% 1734 31%
 89 Sayuri 2015.01.17        A32 4 1654 101 104 36 43% 1703 19%
 90 Tscp_1.8.1.AB_xb         A32 1 1607 118 118 26 50% 1607 31%
 91 Jester_0.84.JA_xb        A32 1 1603 134 124 24 71% 1412 17%
 92 Rocinante 2.0 JA         A32 1 1580 105 104 36 60% 1486 14%
 93 Kitteneitor_060513.JA_xb A32 1 1561 114 117 24 46% 1566 42%
 94 Zurichess                A32 1 1545 151 136 22 75% 1334 14%
 95 Chenard_2014.07.11.JA_xb A32 1 1518 122 129 26 37% 1621 19%
 96 Pulse 1.5-cpp            A32 1 1496 102 100 36 64% 1369 33%
 97 K2 v.071                 A32 1 1463 172 187 18 19% 1847 17%
 98 VIRUTOR CHESS 1.1.1      A32 1 1432 102 102 36 54% 1376 25%
 99 Chess for Android        A32 1 1426 106 106 36 54% 1377 14%
100 Toledo_Nano_2010.JA_xb   A32 1 1397 151 174 18 19% 1692 17%
101 Octochess                A32 4 1385 163 192 18 19% 1694  6%
102 Byak 8.10.14.JA          A32 1 1101 145 181 22 14% 1415  9%
103 Xadreco_5.7.JA_xb        A32 1 1095 146 181 22 11% 1416 14%
104 OliveChess 0.2.7         A32 1  988 194 152 24  0% 1515  0%

Rapidroid test platform:
* GT-N7100 1.7 Ghz quad, 256MB hash: All Android progs,
* Polypad 1010IPS tablet 1.61 Ghz dual, 128MB hash: SP Android progs,
* HTC Diam 528Mhz, 16MB hash: Windows Mobile progs,
* i7 M620 2.67 Ghz dual + Arena 3.5 + 2GB hash: Win64
* iPod Touch 64G @ 600 Mhz to run IOS programs
* DosBox 1.74 to run DOS programs,
* WinVICE 2.24 to run Commodore-64 programs,
* Messtiny UCI adapters or CB-Emu2014 used to emulate Mephisto programs,
* Own books disabled and replaced by 20 ply openings taken from Adam Hair's book,
* Openings selected for max variety, no Q exchange, +0.15 to +0.40 on start,
* Openings played twice with both colors, whenever possible,
* Repeating openings and twin games not allowed between two programs
* Tablebases and pondering off
* Time control: 1/10 or 600+2 or 900+0 or closest valid for both programs.


Android UCI engine update: Sayuri 2015.03.02.JA

Sayuri 2015.01.07 is updated to 2015.03.02 on Windows recently. It's not surprising to see Jim Ablett not skipping the opportunity to catch another update to port to Android. He's simply fast. 

By the way, my sense of transparency about the share of information is forcing me hard to declare the least useful finding of the day which is that Sayuri is specific for being the 2nd lowest rated Android engine after Octochess among to ones which support multi threads as of today. I really couldn't hide that fact anymore...

Send this message to at least ten friends of yours and live the luckiest day of your life or you will soon be checkmated by a chess computer. LoL.

New Sayuri version is available: HERE

New Android XB engine: Joanna 2002.JA

Yet another old engine compiled by Jim Ablett for Android is Joanna 2002. With 1324 ELO on a PC and possibly 1150 on Android, Joanna promises the taste of revenge by granting to human kind a lot of wins versus a chess computer (!).
On my side, if eligible for tournament play, i'll be more than happy to include Joanna in Rapidroid experiment because by the nature of the preferred engine population, i need engines from every ELO segment from the strongest to the weakest one. Old engines are even better for any rating experiment because they can easily become trusted long term anchors as there are no updates to destroy the games already played. Sigh...

Now, your turn: if you wanna knock out a computer in chess, go download and try Joanna: HERE

New Android UCI engine: Protector 1.8.JA

Protector is one of the top chess engines available today. It's written by Reimund Heid from Germany and due to its multi processing compatibility, participates regularly to TCEC events.

According to CCRL, Protector 1.7 is rated 3090 ELO on a PC. Now that we have v1.8 on Android, we may expect 2800 or more in our hands. That's why i simply can't wait to test it and label it "Good for Rapidroid".

However, one trick is worth a little doubt: Jim Ablett recommends Polyglot adapter for Protector to run under Chess for Android, in contrast to Droidfish GUI which can directly use it as a UCI engine.

In deeper detail, Chess for Android needs Polyglot adapter installed as a XB engine and renamed to Protector, then configured via Protector.ini in order to find the real engine binary and manage protocol translations between the engine and CfA.

The presence of a relaying adapter between the GUI and the engine may cause time delay problems in theory but i didn't yet checked whether that leads to time forfeits.

For further info about Polyglot adapter usage with Chess for Android GUI, refer to previous post about Pawny.

Protector 1.8 binary for Android and necessary additional files can be downloaded: HERE

In case you don't like the engine filename i've used, you can change it but in such case, don't forget to modify the ini file which must adress the new names and/or paths. Pay attention to NOT RENAME THE INI FILE because protector.ini is hard coded into provided Polyglot binary. Otherwise, even though the engine starts without problem, it will soon exit with error. Even Protector.ini causes problem because of P.



Android UCI engine update: Pawny 1.0.JA

I remember an old build of Pawny was available for Android before this new build compiled by Jim Ablett. If my memory is good, the previous one was compiled by Jim too, way long ago but i remember that Pawny didn't work for me under Chess for Android, while it played without problem using Fischer clocks under Droidfish.

As complementary information joint to the release, JA states similar observations. So, Droidfish is fine again to run Pawny but Chess for Android, despite its recent 5.0.5 version which now supports Fischer clocks, will still require a trick to run Pawny.

That trick needs Polyglot to be renamed as Pawny and installed as XB engine under Chess for Android. To complete the full setup, we need the provided config file with ini extension, containing the necessary info for Polygot to find and drive the real Pawny engine which must be also installed like an engine under Chess for Android. You never need to import the real engine but the renamed Polyglot binary which is designed to represent it.

The ini file SHOULD NOT BE RENAMED according to JA. The filename Pawny.ini is hard coded into Polyglot binary. If ini renamed, the engine stops working with error on second move.

I will check whether it's possible to just copy the real engine into eng folder of CfA. In any case, installing it is safe as described above.

You see how simple is the procedure! If you are familiar with the method to run UCI engines with Winboard on a PC, you can proceed exactly the same way on Android.

I didn't yet test how good Pawny works that way. You can download all necessary files to give a try: HERE

New Android UCI engine: Murka 3.JA

Murka by Igor Korshunov is a successor to his WildCat, started from scratch and first released in May 2010. The 3rd version hereby ported to Android by Jim Ablett is dated July 2013.

Murka is a Russian common pet name for a cat. I can't tell you how wild is that cat but note that Murka is not a top engine, not a weak player either. Its CCRL rating for PC is 2830.

Download Android compile: HERE

Polyglot adapter brings UCI-2-WB conversion to Android

Once i was wondering why there's no UCI to WB or WB to UCI protocol adapters compiled for Android. It would be useful in many cases where a translation is necessary to make an engine work correctly.
It's good news that Jim Ablett fills the gap now. We can finally call Polyglot adapter to help work a UCI engine like a Winboard engine. That may not remedy but at least bypass compatibility issues.
Furthermore, another adapter which works in reverse direction would be even more useful and would make many XB engines compatible with Droidfish, bearing in mind that Peter Österlund's GUI has some superior features, although engine tournaments aren't still supported.
Therefore, even if Polyglot is highly welcome to our handy devices, we keep waiting for a WB2UCI adapter for Android.

Download Polyglot for Android: HERE

March 6, 2015

New Android UCI engine build: Cinnamon 1.2b.JA

As another mid-range engine, Cinnamon can't take too much priority. However, for anyone interested in enlarging his engine collection, it could be an alternative.

A old build of Cinnamon existed for Android since a long time but my humble experience with it was not succesfull at all. I hope Jim Ablett's new build will change that so that i can add it to RAPIDROID.

Download Cinnamon: HERE

Update 07-Mar-15: Too bad i can't get it to be imported into Chess for Android on Galaxy Note II. It just exits all the time.

New Android XB engine: Soldat III 1.78.JA

Here we go again: Jim Ablett is adding fuel to his compile tank with Soldat III, an engine written by Marco Giusfredi from Italy.

Born in 2002, Soldat is relatively old and plays an average level of chess given that its CCRL rating is 2039.

Download Soldat: HERE

March 4, 2015

R.I.P. Leo Dijksmann

You must have witnessed the rise of winboard chess engines in late 90's to remember exactly how valuable was Leo Dijksmann's contribution to the community.

It would be hard to point a computer chess freak who's never heard of WBEC RIDDERKERK or has never been to take a look at this website for a precise detail of an engine.

We've lost him. Rest in peace and know you won't be forgotten.
His heritage by the way is still in service of the chess lovers at:
http://wbec-ridderkerk.nl

March 3, 2015

New Android XB engine: Trappy Beowulf 2.0.JA

Yes it's trappy. Latest compile by Jim Ablett has ported the "Trappy kind of Beowulf" to Android as an xboard/winboard engine.
Trappy is, as far as i understood, a fork of Beowulf previously ported to Android, originally written by Colin Frayn from UK. Trappy is maintained by Mike Vollmer as an alternative to original Beowulf. The difference is the insertion of "Trappy Minimax" algorithm.

So what is that Trappy Minimax? It says using Iterative Deepening to Identify and Set Traps in Two-Player Games. Um, well, ok. Do it then!

Unfortunately, Beowulf is not in Rapidroid experiment due to its tendancy to use more time than specified. I hope Trappy does better and gets in the ring.

Download Trappy Beowulf: HERE

March 1, 2015

Thank you for a very enjoyable game

The post title refers to the words of a computer, seen in the header image of my blog which is taken from 2001: A SPACE ODYSSEY, showing Frank Poole losing a chess game to HAL9000, the supercomputer in charge of almost everything in the spacecraft on the way to... hmm well let's say to beyond the infinite, as seen at the end of the movie.
There are million words to tell about the movie itself which is for me a very deep combination of multiple stories but what i want to remember now is the game itself which in fact is an early prediction of the chess computers taking over humans in that gigantic puzzle.

While the movie is a science-fiction, the game Poole loses to HAL is a very old one, played by true human players in early 1900s. It's kind of a short and sweet hit by black after the way too greedy Qxa8 quickly punished.

Here's the position from where a good 12.d4 would be enough to keep the balance:
11... Nxe5 just before 12.Qxa8??

The full game:


New Android XB engine: Sunsetter C10.JA

Sunsetter C10 belongs to Georg v. Zimmermann from Germany. In parallel to classical chess, the author is also focused on other variants as well, mainly Crazyhouse chess for a more difficult task, given that this variant needs much more nodes to achieve the same search depth.

Anyway, we Androiders, are not done yet with classical chess!

For all those who need another classical chess engine on Android, we have now Sunsetter compiled by Jim Ablett available for download in my box repository: HERE

According to CCRL, Windows version of Sunsetter plays at 2146 ELO. Should be around 1950 on Android.