This GUI is capable of playing round-robin. Just select as many engines as you wish when you setup the tourney.
On my device senpai works for both ponder on and ponder off.
Arasan works on ponder off, but not in ponder on.
Chess for Android v5.0: Time Control
Moderator: Ras
-
Ferdy
- Posts: 4851
- Joined: Sun Aug 10, 2008 3:15 pm
- Location: Philippines
-
Sedat Canbaz
- Posts: 3018
- Joined: Thu Mar 09, 2006 11:58 am
- Location: Antalya/Turkey
Re: Chess for Android v5.0: Time Control
Ferdy wrote:This GUI is capable of playing round-robin. Just select as many engines as you wish when you setup the tourney.
On my device senpai works for both ponder on and ponder off.
Arasan works on ponder off, but not in ponder on.

Of course Senpai works on my device too,
But the most important question: HOW ??))
See the above picture please ...
And it looks like there is something wrong,
e.g Senpai oftenly forfeits on time and does not go into the next rounds....
About your rest comments,
Thanks, I noticed it... yes the GUI supports Round-Robin system
But my wish is simply is that:
We need more options, e.g: Cycles
Otherwise (depending on the number of database games) many games are playing ....
Note that I am interested testing the Android engines:
All games from external book
And one of the most important issues - its will be great if we will see:
Cycle option and After each game to see different participants...otherwise it's too boring to see same opponent vs...
For Example,
Arena GUI, ShredderClassic, ChessBase etc... have 'Cycle' options plus after each game we see different opponents...
And I will be very happy too, if we will see Gauntlet system too
In this way, we can test various android engines in a short time...
About Arasan,
Off...even the Arasan 17.4 Windows versions is buggy too
Best,
Sedat
-
Sedat Canbaz
- Posts: 3018
- Joined: Thu Mar 09, 2006 11:58 am
- Location: Antalya/Turkey
Re: Chess for Android v5.0: Time Control
One thing more (for more various openings),
It will be excellent news, if there will be another extra testing option:
- The starting opening positions to be loaded randomly!
Thanks in advance )
Sedat
It will be excellent news, if there will be another extra testing option:
- The starting opening positions to be loaded randomly!
Thanks in advance )
Sedat
-
Gurcan Uckardes
- Posts: 196
- Joined: Wed Oct 29, 2014 12:42 am
Re: Chess for Android v5.0: Time Control
Many engines get flagged. Especially the weaker ones. Grrr.
I'm Currently testing at 3m/game. Will report the result in a few days for all available engines. Fixed time per move will be hard to abandon on wide range rating list experiments.
I'm Currently testing at 3m/game. Will report the result in a few days for all available engines. Fixed time per move will be hard to abandon on wide range rating list experiments.
My blog for Android users: http://chesstroid.blogspot.com
-
abik
- Posts: 823
- Joined: Fri Dec 01, 2006 10:46 pm
- Location: Mountain View, CA, USA
- Full name: Aart Bik
Re: Chess for Android v5.0: Time Control
I can come up with several explanations, feel free to add your own ideas.Gurcan Uckardes wrote:Many engines get flagged. Especially the weaker ones. Grrr.
I'm Currently testing at 3m/game. Will report the result in a few days for all available engines. Fixed time per move will be hard to abandon on wide range rating list experiments.
- When using time control for the whole game, some engines simply run out of time toward the end of longer games due to their time management; they usually do much better using time control for a fixed number of moves.
- Some engines simply use the full amount of time given not accounting for any overhead between the GUI and engine; having said that, I found that this does not happen for the stable engines, including e.g. the Xboard engine Crafty (for Xboard the time control is only sent at start; for UCI it gets an update every move so one would expect less issues).
- Perhaps the time management implementation is broken for some engines when compiled for ARM and/or Android.
-
Ferdy
- Posts: 4851
- Joined: Sun Aug 10, 2008 3:15 pm
- Location: Philippines
Re: Chess for Android v5.0: Time Control
I don't think a TC like 3min/game is an ideal TC for testing engines, this is only good for time stress test where the engine should at least survive 200 moves or so, why not give it some breathing room like 1 sec increment. In your posted screen shot, senpai was already losing, it was just trying to use more time in a losing position perhaps. Another thing to consider is how much hash value you give to these engines, the device must have enough left to process other things. It is not only the hash that should be looked at there are engines that uses other memory, example Komodo has Table Memory. CPU usage has to be monitored too. How many cores do you allocate for each engine? there are lot of things to consider here.Sedat Canbaz wrote:Ferdy wrote:This GUI is capable of playing round-robin. Just select as many engines as you wish when you setup the tourney.
On my device senpai works for both ponder on and ponder off.
Arasan works on ponder off, but not in ponder on.
Of course Senpai works on my device too,
But the most important question: HOW ??))
See the above picture please ...
And it looks like there is something wrong,
e.g Senpai oftenly forfeits on time and does not go into the next rounds....
I tried 2 minutes/game, komodo 8 vs senpai, 94/100 games so far ponder on, single thread, no time forfeits, or no flagged games, the tourney has no problem. I only set 16 mb hash for each engine. My device has 4 cores with 1GB RAM only, with 8GB internal still with more space and an external of 32 GB also with some space left. This is on Android kitkat.
-
Gurcan Uckardes
- Posts: 196
- Joined: Wed Oct 29, 2014 12:42 am
Re: Chess for Android v5.0: Time Control
3 mins is a little bit low for Android because of lower knps levels. I recomment at least 1 M nodes per move average. Rating lists may distort near blitz levels.
But i still use severe levels like 2+0 for testing the functioning.
Almost all xb engines match the fact Aart mentioned above. And Senpai 1.0.JA is flagged in long games.
Would it be a solution to wait 0.1 secs per move by the gui, kind of a overhead? Thus, CfA waits such amount before restarting the clock. Indeed, total game time increases 15-20 seconds. I think it's bearable and better than forfeits.
Arena has this option. Komodo Android too.
But i still use severe levels like 2+0 for testing the functioning.
Almost all xb engines match the fact Aart mentioned above. And Senpai 1.0.JA is flagged in long games.
Would it be a solution to wait 0.1 secs per move by the gui, kind of a overhead? Thus, CfA waits such amount before restarting the clock. Indeed, total game time increases 15-20 seconds. I think it's bearable and better than forfeits.
Arena has this option. Komodo Android too.
My blog for Android users: http://chesstroid.blogspot.com
-
Ferdy
- Posts: 4851
- Joined: Sun Aug 10, 2008 3:15 pm
- Location: Philippines
Re: Chess for Android v5.0: Time Control
Gurcan Uckardes wrote:3 mins is a little bit low for Android because of lower knps levels. I recomment at least 1 M nodes per move average. Rating lists may distort near blitz levels.
But i still use severe levels like 2+0 for testing the functioning.
This is fine.
The senpai that I have is not JA.Almost all xb engines match the fact Aart mentioned above. And Senpai 1.0.JA is flagged in long games.
Would it be a solution to wait 0.1 secs per move by the gui, kind of a overhead? Thus, CfA waits such amount before restarting the clock. Indeed, total game time increases 15-20 seconds. I think it's bearable and better than forfeits.
Arena has this option. Komodo Android too.
-
Sedat Canbaz
- Posts: 3018
- Joined: Thu Mar 09, 2006 11:58 am
- Location: Antalya/Turkey
Re: Chess for Android v5.0: Time Control
Just a moment,Ferdy wrote:I don't think a TC like 3min/game is an ideal TC for testing engines, this is only good for time stress test where the engine should at least survive 200 moves or so, why not give it some breathing room like 1 sec increment. In your posted screen shot, senpai was already losing, it was just trying to use more time in a losing position perhaps. Another thing to consider is how much hash value you give to these engines, the device must have enough left to process other things. It is not only the hash that should be looked at there are engines that uses other memory, example Komodo has Table Memory. CPU usage has to be monitored too. How many cores do you allocate for each engine? there are lot of things to consider here.Sedat Canbaz wrote:Ferdy wrote:This GUI is capable of playing round-robin. Just select as many engines as you wish when you setup the tourney.
On my device senpai works for both ponder on and ponder off.
Arasan works on ponder off, but not in ponder on.
Of course Senpai works on my device too,
But the most important question: HOW ??))
See the above picture please ...
And it looks like there is something wrong,
e.g Senpai oftenly forfeits on time and does not go into the next rounds....
I tried 2 minutes/game, komodo 8 vs senpai, 94/100 games so far ponder on, single thread, no time forfeits, or no flagged games, the tourney has no problem. I only set 16 mb hash for each engine. My device has 4 cores with 1GB RAM only, with 8GB internal still with more space and an external of 32 GB also with some space left. This is on Android kitkat.
My published picture (where Senpai lost on time) is at 3 min +1 sec
And since 2 days I started testing the Android engines at 3m+1s
And as we see, even at 3m+1s Senpai again forfeits on time...
Simply Senpai android seems to be buggy one, and you can not blame the blitz time control !)
Plus, how are you sure that Senpai never loose on time on your device ?
Did you check each played game played so far ? )
For example,
I am using Chessbase GUI for determining which games are lost on time
Otherwise without checking....
You can not be sure about games played at fast time control...
Even if you are in front of your device (during active match) you can miss some of the games...
BTW, can I see/download your games played by Senpai ?
-
Ferdy
- Posts: 4851
- Joined: Sun Aug 10, 2008 3:15 pm
- Location: Philippines
Re: Chess for Android v5.0: Time Control
I assume you use 3 minutes only because of your previous post saying...Sedat Canbaz wrote:Just a moment,Ferdy wrote:I don't think a TC like 3min/game is an ideal TC for testing engines, this is only good for time stress test where the engine should at least survive 200 moves or so, why not give it some breathing room like 1 sec increment. In your posted screen shot, senpai was already losing, it was just trying to use more time in a losing position perhaps. Another thing to consider is how much hash value you give to these engines, the device must have enough left to process other things. It is not only the hash that should be looked at there are engines that uses other memory, example Komodo has Table Memory. CPU usage has to be monitored too. How many cores do you allocate for each engine? there are lot of things to consider here.Sedat Canbaz wrote:Ferdy wrote:This GUI is capable of playing round-robin. Just select as many engines as you wish when you setup the tourney.
On my device senpai works for both ponder on and ponder off.
Arasan works on ponder off, but not in ponder on.
Of course Senpai works on my device too,
But the most important question: HOW ??))
See the above picture please ...
And it looks like there is something wrong,
e.g Senpai oftenly forfeits on time and does not go into the next rounds....
I tried 2 minutes/game, komodo 8 vs senpai, 94/100 games so far ponder on, single thread, no time forfeits, or no flagged games, the tourney has no problem. I only set 16 mb hash for each engine. My device has 4 cores with 1GB RAM only, with 8GB internal still with more space and an external of 32 GB also with some space left. This is on Android kitkat.
My published picture (where Senpai lost on time) is at 3 min +1 sec
And since 2 days I started testing the Android engines at 3m+1s
And as we see, even at 3m+1s Senpai again forfeits on time...
Simply Senpai android seems to be buggy one, and you can not blame the blitz time control !)
Sedat Canbaz wrote:Btw, Senpai android at TC 3 minutes does not want to play...
Arasan 17.4 android seems to be buggy too...
I use HCE, and notepad++ to verify the games and its results.Sedat Canbaz wrote: Plus, how are you sure that Senpai never loose on time on your device ?
Did you check each played game played so far ? )
For example,
I am using Chessbase GUI for determining which games are lost on time
Otherwise without checking....
You can not be sure about games played at fast time control...
Even if you are in front of your device (during active match) you can miss some of the games...
BTW, can I see/download your games played by Senpai ?
Download the games here.
http://www.mediafire.com/download/o1sb7 ... -games.rar
Another note, I am not using the ja compile of senpai, we could be different here.