I'm sure fishers will be happy to install the latest development version of Stockfish ported to Android. Thanks must be given to Peter Österlund for compiling it.
Based on ongoing blitz gauntlets (not yet shared here), even 2016.05.28 is way stronger than Stockfish 7 which came with Droidfish 1.60.
I guess the August release of Rapidroid will include this development version of Stockfish versus Komodo 10. Rounds are currently being replayed on my Exynos' to reveal this. And be sure that about tthe poor boys are about to melt in my refrigerator!
GET Stockfish 2016.07.16 for arm5 32-bit nopie
GET Stockfish 2016.07.16 for arm5 32-bit pie
GET Stockfish 2016.07.16 for arm7 32-bit nopie
GET Stockfish 2016.07.16 for arm7 32-bit pie
GET Stockfish 2016.07.16 for arm8 64-bit
GET Stockfish 2016.07.16 for x86 32-bit nopie
GET Stockfish 2016.07.16 for x86 32-bit pie
GET Stockfish 2016.07.16 for x86 64-bit
31 comments:
Hi! Stockfish 120616 is stronger than stockfish 2016.07.16. I made 28 gmae between stockfish120616 and stockfish2016.07.16 and result is= STOCKFISH120616-17.5 and stockfish2016.07.16-10.5. So i say that this update is not good.
28 the party is small,the test error tends to infinity.Even after playing 200 parties,the error would be about +-40 Elo.Because so many parties you can't judge the power of the engine.Also,what is the time control,the number of cores?
I made all game on android quad-core 1.3GHz on hawk chess server.
I don't have time to choose between builds for Rapidroid. The fair choice should be the closest development version to Komodo 10.
By the way, i will surely report a gauntlet results later on. Based on 612 Tcec-7 games (306 openings x 2) to be played on Rockchip 3188 running 4 cores @ 1ghz but with 300+1 time control. It's about 5.5 times less nodes than Rapidroid but gives an idea.
Алексей Чернаков is the latest sf dv 16-7-16 weaker than 12-6-16?
do you have any mtch result?
I still spent 90 parties,control 2'm+1's.Preliminary result:
Stockfish 120616=50.5/90.
Stockfish 160716=39.5/90.Stockfish Stockfish 120616 160716 slower by 5-10%.Worse gaining depth.
hmm seems the latest sf is weaker strange..
Stockfish120616 46/100
komodo10 54/100
Time control?
Time control 2m +1s
fantastic.
Qualcomm snapdragon 805 @ 2.7GHz, 3Gig RAM
Syzygy tablebases 5 pieces
Perfect 2015 x up to 8 moves
Stockfish 20160626 - 49/100
Stockfish 20160716 - 51/100
I've finished playing arpund 80 games out of 300+ for Rapidroid. Komodo 10 surprisingly chases SF160716 extremely closer than ever. I see K10 got stronger against others while SF didn't boost as much as it does vs SF7. I still wait for the a gauntlet between SF devs. I'm not yet sure about a refression. Peter's builds used to perform well all the time.
Nathanael, thanks for the feedback. Your result reduces the doubt. To me there's no reason for the last dev weakens, especilly with Peter's build.
Qualcomm snapdragon 805 @ 2.7GHz, 3Gig RAM
Syzygy tablebases 5 pieces
Perfect 2015 x up to 8 moves
Each side 1 sec per move until the game ends
Stockfish 20160612 - 48.5/100
Stockfish 20160716 - 51.5/100
Chess for Android: https://www.dropbox.com/s/9o90f0t06910538/Screenshot_2016-07-20-16-20-59.png?dl=0
but the latest dv P.O is slower in my cheap phone:16-7:90kn vs 16-6:102kn 1 core arm 1 ghz
Stockfish20160716 40.5/100
komodo10 59.5/100
Time control 2m+1s game played on android device Quad-core1.3GHz.
Lenovo, 4 cores, 1Gb RAM, 1.3GHz, time 2m+1sec, syzysy 5 man, ponder OF.
Stockfish 120616-51,5
Komodo 10-48,5
Komodo10 is not good when i set threds=4 and hash=32 for both engine then komodo badly crushed by stockfish120616
stockfish66/100
komodo10 34/100
Second round->Lenovo, 4 cores, 1Gb RAM, 1.3GHz, time 2m+1sec, syzysy 5 man, ponder ON.
Stockfish 120616-53
Komodo 10-47
How to set syzysy.
It doesn't matter how many hash I will use, 32 or 64 for each engine. Komodo 10 is not able to play with 100% of strength, on devices with 1GB ram.
Same thing with ST, on my device best version is 120616, P.O. 2016.07.16 coming in second place, with small difference on kN/s , about 60-70 kns slower then 120616.
The whole point is, faster device, better results for ST by P.O and K10!
Rahul, in Chess for android (C4a), first you need to clear options for Stok and K10.
Open C4a ->long press on the screen ->UCI and XBoard->Clear Options->choose engines (mark them) -> press back arrow to return to the C4a, after that again long press on the screen, UCI and XBoard ->Import Engine ->choose Stockfish, in the engine menu, set hash, cores, and down below,if you have syzygy on your device, just write syzygy path, for example /mnt/sdcard/syzygy -> pres Done. That's all.
Rahul, i forgot to mention, you need to do the same thing for Komodo. Create path for syzygy, as you l'll do for Stockfish.
Stockfish 2016.07.16 for arm7 32-bit nopie
Stockfish 2016.07.16 for arm7 32-bit pie
What means "pie" and "nopie"?
Thanks!
Rahul, one last thing, if the path for syzygy is correct, you will see a short message from engine-> found 146 tablebases.
With this control,syzygy is better not to use,as it will dramatically reduce the performance of the engine.The engine will dig into the tables,rather than rely options,and time is short.
Use pie with Android Lollipop and Marshmallow. Nopie doesn't work with these. For Android older than 4.1, use no pie. 4.1 to 4 4: You can use both.
Please can anybody explain me what means Pie or nopie
Iam searching but nothing found now i knew that my Handy has an ARM 7
Chipset Qualcomm Snapdragon 800
CPU Quad-core 2.2 GHz Krait 400
GPU Adreno 330
What what is the right version for my tablet
You need Arm7 pie, asauming your Android is 32bit.
Post a Comment