Rating close to fide elo

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
pedrox
Posts: 1056
Joined: Fri Mar 10, 2006 6:07 am
Location: Basque Country (Spain)

Re: laskos

Post by pedrox »

To play with openings book, it should be changed the danasah.txt file:

ownbook true


If you set up danasah to play with 2000 Elo points, you should play with the same force regardless of whether you play on a computer with 3.4 GHz or a computer with 2.4 GHz, as the strength depends largely on the nps.

Between 1400 and 2100 points I use the formula:
nps_elo = (pow(2,((double)elo)/100.0) / 100) * ((2200 - elo) / 100);

With elo = 2000 --> nps = 20.971

Danasah on a 2.4 GHz processor usually makes about 500,000 nps, to play with 2000 points we would need only 20,971 nps, so much of the time the engine is in sleep mode. So the engine will play with the same force regardless of the processor, to play with 2000 elo points, the cpu will always have to be strong enough to produce those 20,791 nps.

In addition to the number of nps I apply a random factor to the evaluation.

All this is in reference to the type_elo engine, for the other types (computer and human) I weaken the force even more.

If you use:

Code: Select all

limitstrength true 
elo 2000 
elo_type human 
style 3 
diff_engine 0 
diff_computer_engine 466 
diff_human_computer 70 
easylevel 0 
ownbook true 
tourbook ./book/none.bin 
mainbook ./book/super.bin 
hash 128 
egbb_path 
egbb_men 4 
egbb_cache_size 4 
egbb_load_type 1 
noise 0 
castling KQkq
If you see that danasah plays very loose and was about 400 points lower than your engine playing 2000 ELO FIDE points, then the solution to play 400 points stronger would be to reduce the parameter diff_computer_engine value for those 400 points in dasanah.txt, for example:

diff_computer_engine 66

Maybe my 466 value is too high, but I've seen some test playing with a elo of about 2000 and with type_elo computer then danasah plays fairly matched to the Mephisto Rome machine with a 30 minute control time for the game.

I have a suggestion for you, check if the result you have with your engine and the Mephisto 3 minutes plus 2 seconds is also maintained with the time control of 30 minutes per game or 30 seconds per move.
Ferdy
Posts: 4848
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: laskos

Post by Ferdy »

pedrox wrote:To play with openings book, it should be changed the danasah.txt file:

ownbook true


If you set up danasah to play with 2000 Elo points, you should play with the same force regardless of whether you play on a computer with 3.4 GHz or a computer with 2.4 GHz, as the strength depends largely on the nps.

Between 1400 and 2100 points I use the formula:
nps_elo = (pow(2,((double)elo)/100.0) / 100) * ((2200 - elo) / 100);

With elo = 2000 --> nps = 20.971

Danasah on a 2.4 GHz processor usually makes about 500,000 nps, to play with 2000 points we would need only 20,971 nps, so much of the time the engine is in sleep mode. So the engine will play with the same force regardless of the processor, to play with 2000 elo points, the cpu will always have to be strong enough to produce those 20,791 nps.

In addition to the number of nps I apply a random factor to the evaluation.

All this is in reference to the type_elo engine, for the other types (computer and human) I weaken the force even more.

If you use:

Code: Select all

limitstrength true 
elo 2000 
elo_type human 
style 3 
diff_engine 0 
diff_computer_engine 466 
diff_human_computer 70 
easylevel 0 
ownbook true 
tourbook ./book/none.bin 
mainbook ./book/super.bin 
hash 128 
egbb_path 
egbb_men 4 
egbb_cache_size 4 
egbb_load_type 1 
noise 0 
castling KQkq
If you see that danasah plays very loose and was about 400 points lower than your engine playing 2000 ELO FIDE points, then the solution to play 400 points stronger would be to reduce the parameter diff_computer_engine value for those 400 points in dasanah.txt, for example:

diff_computer_engine 66

Maybe my 466 value is too high, but I've seen some test playing with a elo of about 2000 and with type_elo computer then danasah plays fairly matched to the Mephisto Rome machine with a 30 minute control time for the game.

I have a suggestion for you, check if the result you have with your engine and the Mephisto 3 minutes plus 2 seconds is also maintained with the time control of 30 minutes per game or 30 seconds per move.
TC 30 minutes for the whole game, silver50 test suite, opening is played twice, results so far with 6 draws.

Code: Select all

Score of CDrill 1800 vs Mephisto Amsterdam: 8 - 8 - 6 [0.500]
Elo difference: 0.00 +/- 129.29

22 of 100 games finished.
Ferdy
Posts: 4848
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

CDrill - Mephisto Amsterdam Rapid

Post by Ferdy »

I ended the test, this is the last result.

Code: Select all

Score of CDrill 1800 vs Mephisto Amsterdam: 11 - 11 - 6 [0.500]
Elo difference: 0.00 +/- 118.27

29 of 100 games finished.
User avatar
pedrox
Posts: 1056
Joined: Fri Mar 10, 2006 6:07 am
Location: Basque Country (Spain)

Re: CDrill - Mephisto Amsterdam Rapid

Post by pedrox »

Thanks for the test, it seems that the engine plays the same in blitz and in rapid time against Mephisto.

When I have a little time I will test your engine and try to check if my configuration (why?) as a computer and human is really weak.
User avatar
pedrox
Posts: 1056
Joined: Fri Mar 10, 2006 6:07 am
Location: Basque Country (Spain)

Re: CDrill - Mephisto Amsterdam Rapid

Post by pedrox »

Hi Ferdinand,

I am trying to understand why my engine looks similar to the Mephisto Amsterdam when I set it up as 2000 human and then it seems to lose clearly against your engine.

I have doubts about the confrontation of your engine against Mephisto and if the configuration is the same as mine.

I imagine that to try the Mephisto Amsterdam you have used the Mess Mephisto package with Winboard.

Here is a part of the debug:

Code: Select all

StartChildProcess (dir="..") "./messtiny.exe" amsterd -nommunlimited
1154 >second: xboard
protover 2
1156 <second: Mess Mephisto Version: 0.138.d Winboard (Jul  5 2010)
1157 <second: 
1157 <second: Mephisto Amsterdam
1157 <second: 
1157 <second: Emulator org. clock : 12000000
1157 <second: Emulator curr.clock : 12000000
1158 <second: Speed factor clock  : 1.00
1158 <second: 
1158 <second: OS ticks_per_second : 3237194
1158 <second: 
1158 <second: Supported Levels:
1158 <second: 
1158 <second: Level 0: 3   seconds/move   -> st 3
1159 <second: Level 1: 5   seconds/move   -> st 5
1159 <second: Level 2: 10  seconds/move   -> st 10
1159 <second: Level 3: 20  seconds/move   -> st 20
1159 <second: Level 4: 60  seconds/move   -> st 60
1160 <second: Level 5: 120 seconds/move   -> st 120
1160 <second: Level 6: 40 move in 2 hours -> level 40 120 0
1160 <second: 
1160 <second: PLAy
1160 <first : pong 1
1553 <second: feature sigint=0 ping=1 setboard=1 color=0 done=1  myname="Mephisto Amsterdam" 
1553 >second: accepted sigint
1553 >second: accepted ping
1553 >second: accepted setboard
1553 >second: rejected color
1554 >second: accepted done
1554 >second: accepted myname
1573 >second: new
random
1573 >second: st 5
1573 >second: post
1574 >second: hard
1574 >second: easy
1574 >second: ping 1
1574 >second: force
1574 >first : computer
1574 >second: computer
TC string = '+40/5'
mps=40 tc=300000 inc=0
TC string = '+40/5'
mps=40 tc=300000 inc=0
book hit = (NULL)
1575 >first : go
When setting the engine I used the -nommunlimited parameter for the engine to use the original 12 MHz frequency.

And another important thing is that the Mephisto Amsterdam only has 7 levels of time control so you should use one of the available or otherwise the machine plays the default level.

For example I have tried level 1 -> st 5

If the -nommunlimited option is not available then the Mephisto will play according to the strength of the computer.

Code: Select all

4 Jan 2017 20:02:16 - Mess Mephisto Version: 0.138.d Winboard (Jul  5 2010)

04 Jan 2017 20:02:16 - Mephisto Amsterdam

04 Jan 2017 20:02:16 - Display: PLAy
04 Jan 2017 20:02:16 - Emulator org. clock : 12000000
04 Jan 2017 20:02:16 - Emulator curr.clock : 3000000
04 Jan 2017 20:02:16 - Speed factor clock  : 0.25

04 Jan 2017 20:02:16 - OS ticks_per_second : 3237194

04 Jan 2017 20:02:16 - Time in ms for 1 sec: 11
04 Jan 2017 20:02:17 - Speed factor time   : 90.91

04 Jan 2017 20:02:17 - Speed factor total  : 22.73

04 Jan 2017 20:02:17 - ReferenceTimePerSec : 94 (This system has: 11)
04 Jan 2017 20:02:17 - Factor time corr.   : 0.12
04 Jan 2017 20:02:17 - g_tc_delay (ms)     : 234

04 Jan 2017 20:02:17 - GUI    Input : xboard
04 Jan 2017 20:02:17 - GUI    Input : protover 2
04 Jan 2017 20:02:17 - ENGINE Output: feature sigint=0 ping=1 setboard=1 color=0 done=1  myname="Mephisto Amsterdam" 
04 Jan 2017 20:02:17 - GUI    Input : accepted sigint
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : accepted ping
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : accepted setboard
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : rejected color
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : accepted done
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : accepted myname
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : new
04 Jan 2017 20:02:17 - Softreset
04 Jan 2017 20:02:17 - GUI    Input : random
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:18 - GUI    Input : level 40 5 0
04 Jan 2017 20:02:18 - g_cmd: l9s
04 Jan 2017 20:02:18 - Display: LE 2
04 Jan 2017 20:02:18 - Display: LE 9
04 Jan 2017 20:02:18 - Display: PLAy
04 Jan 2017 20:02:18 - GUI    Input : post
04 Jan 2017 20:02:18 - g_cmd: 
04 Jan 2017 20:02:18 - GUI    Input : hard
04 Jan 2017 20:02:18 - g_cmd: 
04 Jan 2017 20:02:18 - GUI    Input : easy
04 Jan 2017 20:02:18 - g_cmd: 
04 Jan 2017 20:02:18 - GUI    Input : ping 1
04 Jan 2017 20:02:18 - ENGINE Output: pong 1
04 Jan 2017 20:02:18 - g_cmd: 
04 Jan 2017 20:02:23 - GUI    Input : time 30000
04 Jan 2017 20:02:23 - g_cmd: 
04 Jan 2017 20:02:23 - GUI    Input : otim 30000
04 Jan 2017 20:02:23 - g_cmd: 
04 Jan 2017 20:02:24 - GUI    Input : e2e4
04 Jan 2017 20:02:24 - xcmd_force_mode: 0
Speed factor total : 22.73

In this case the Mephisto plays with 22.73 times faster than the original on my computer, it will even be faster on your computer than it is faster than mine.

Could it be the case that you have not used that parameter (-nommunlimited)? And that even the time control was not one of the original 7?
Ferdy
Posts: 4848
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: CDrill - Mephisto Amsterdam Rapid

Post by Ferdy »

pedrox wrote:Hi Ferdinand,

I am trying to understand why my engine looks similar to the Mephisto Amsterdam when I set it up as 2000 human and then it seems to lose clearly against your engine.

I have doubts about the confrontation of your engine against Mephisto and if the configuration is the same as mine.

I imagine that to try the Mephisto Amsterdam you have used the Mess Mephisto package with Winboard.

Here is a part of the debug:

Code: Select all

StartChildProcess (dir="..") "./messtiny.exe" amsterd -nommunlimited
1154 >second: xboard
protover 2
1156 <second: Mess Mephisto Version: 0.138.d Winboard (Jul  5 2010)
1157 <second: 
1157 <second: Mephisto Amsterdam
1157 <second: 
1157 <second: Emulator org. clock : 12000000
1157 <second: Emulator curr.clock : 12000000
1158 <second: Speed factor clock  : 1.00
1158 <second: 
1158 <second: OS ticks_per_second : 3237194
1158 <second: 
1158 <second: Supported Levels:
1158 <second: 
1158 <second: Level 0: 3   seconds/move   -> st 3
1159 <second: Level 1: 5   seconds/move   -> st 5
1159 <second: Level 2: 10  seconds/move   -> st 10
1159 <second: Level 3: 20  seconds/move   -> st 20
1159 <second: Level 4: 60  seconds/move   -> st 60
1160 <second: Level 5: 120 seconds/move   -> st 120
1160 <second: Level 6: 40 move in 2 hours -> level 40 120 0
1160 <second: 
1160 <second: PLAy
1160 <first : pong 1
1553 <second: feature sigint=0 ping=1 setboard=1 color=0 done=1  myname="Mephisto Amsterdam" 
1553 >second: accepted sigint
1553 >second: accepted ping
1553 >second: accepted setboard
1553 >second: rejected color
1554 >second: accepted done
1554 >second: accepted myname
1573 >second: new
random
1573 >second: st 5
1573 >second: post
1574 >second: hard
1574 >second: easy
1574 >second: ping 1
1574 >second: force
1574 >first : computer
1574 >second: computer
TC string = '+40/5'
mps=40 tc=300000 inc=0
TC string = '+40/5'
mps=40 tc=300000 inc=0
book hit = (NULL)
1575 >first : go
When setting the engine I used the -nommunlimited parameter for the engine to use the original 12 MHz frequency.

And another important thing is that the Mephisto Amsterdam only has 7 levels of time control so you should use one of the available or otherwise the machine plays the default level.

For example I have tried level 1 -> st 5

If the -nommunlimited option is not available then the Mephisto will play according to the strength of the computer.

Code: Select all

4 Jan 2017 20:02:16 - Mess Mephisto Version: 0.138.d Winboard (Jul  5 2010)

04 Jan 2017 20:02:16 - Mephisto Amsterdam

04 Jan 2017 20:02:16 - Display: PLAy
04 Jan 2017 20:02:16 - Emulator org. clock : 12000000
04 Jan 2017 20:02:16 - Emulator curr.clock : 3000000
04 Jan 2017 20:02:16 - Speed factor clock  : 0.25

04 Jan 2017 20:02:16 - OS ticks_per_second : 3237194

04 Jan 2017 20:02:16 - Time in ms for 1 sec: 11
04 Jan 2017 20:02:17 - Speed factor time   : 90.91

04 Jan 2017 20:02:17 - Speed factor total  : 22.73

04 Jan 2017 20:02:17 - ReferenceTimePerSec : 94 (This system has: 11)
04 Jan 2017 20:02:17 - Factor time corr.   : 0.12
04 Jan 2017 20:02:17 - g_tc_delay (ms)     : 234

04 Jan 2017 20:02:17 - GUI    Input : xboard
04 Jan 2017 20:02:17 - GUI    Input : protover 2
04 Jan 2017 20:02:17 - ENGINE Output: feature sigint=0 ping=1 setboard=1 color=0 done=1  myname="Mephisto Amsterdam" 
04 Jan 2017 20:02:17 - GUI    Input : accepted sigint
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : accepted ping
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : accepted setboard
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : rejected color
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : accepted done
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : accepted myname
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:17 - GUI    Input : new
04 Jan 2017 20:02:17 - Softreset
04 Jan 2017 20:02:17 - GUI    Input : random
04 Jan 2017 20:02:17 - g_cmd: 
04 Jan 2017 20:02:18 - GUI    Input : level 40 5 0
04 Jan 2017 20:02:18 - g_cmd: l9s
04 Jan 2017 20:02:18 - Display: LE 2
04 Jan 2017 20:02:18 - Display: LE 9
04 Jan 2017 20:02:18 - Display: PLAy
04 Jan 2017 20:02:18 - GUI    Input : post
04 Jan 2017 20:02:18 - g_cmd: 
04 Jan 2017 20:02:18 - GUI    Input : hard
04 Jan 2017 20:02:18 - g_cmd: 
04 Jan 2017 20:02:18 - GUI    Input : easy
04 Jan 2017 20:02:18 - g_cmd: 
04 Jan 2017 20:02:18 - GUI    Input : ping 1
04 Jan 2017 20:02:18 - ENGINE Output: pong 1
04 Jan 2017 20:02:18 - g_cmd: 
04 Jan 2017 20:02:23 - GUI    Input : time 30000
04 Jan 2017 20:02:23 - g_cmd: 
04 Jan 2017 20:02:23 - GUI    Input : otim 30000
04 Jan 2017 20:02:23 - g_cmd: 
04 Jan 2017 20:02:24 - GUI    Input : e2e4
04 Jan 2017 20:02:24 - xcmd_force_mode: 0
Speed factor total : 22.73

In this case the Mephisto plays with 22.73 times faster than the original on my computer, it will even be faster on your computer than it is faster than mine.

Could it be the case that you have not used that parameter (-nommunlimited)? And that even the time control was not one of the original 7?
I am using the uci version from Ed's site.
http://rebel13.nl/dedicated/dedicated%20as%20uci.html
User avatar
pedrox
Posts: 1056
Joined: Fri Mar 10, 2006 6:07 am
Location: Basque Country (Spain)

Re: CDrill - Mephisto Amsterdam Rapid

Post by pedrox »

To do the emulation of Mephisto machines use the file "messtiny.exe". Actually messtiny is compatible with the xboard protocol. In this package for ease of use has been converted to the UCI protocol engines simply by using the "Wb2uci.exe" converter.

wb2uci.eng for Mephisto Amsterdam

Code: Select all

[ENGINE]
Name=Mephisto Amsterdam
Author=Richard Lang
Filename=Amsterdam-UCI.exe

[OPTIONS]
Program=..\messtiny.exe amsterd -mmlog
The Mephisto amsterd is called by default in this configuration without the parameter -nommunlimited, if you have used it without this parameter it is possible that the Mephisto Amsterdam has played with all the force of your computer and not with the original frequency of 12 MHz, that should be checked in the log file it generates.

On your computer this could assume that the Mephisto played with a frequency maybe 32 times the original. And that could mean a lot of points. If we win 35 Elo points every time we double that could be about 175 points, 250 points if the Mephisto wins 50 points when doubling, even 350 points if the Mephisto wins 70 points when doubling.

On the other hand, it is possible that my engine when playing as a human needs more time for the regulation to be better, I think I had discovered it even though I had forgotten it over time, I think it needs at least 15 (30) seconds per game or games 15 (30) minutes, otherwise play loose.

I'm going to play some games against Mephisto Amsterdam with time control st5 and st20, I think in the second control will do much better.
Ferdy
Posts: 4848
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: CDrill - Mephisto Amsterdam Rapid

Post by Ferdy »

pedrox wrote:To do the emulation of Mephisto machines use the file "messtiny.exe". Actually messtiny is compatible with the xboard protocol. In this package for ease of use has been converted to the UCI protocol engines simply by using the "Wb2uci.exe" converter.

wb2uci.eng for Mephisto Amsterdam

Code: Select all

[ENGINE]
Name=Mephisto Amsterdam
Author=Richard Lang
Filename=Amsterdam-UCI.exe

[OPTIONS]
Program=..\messtiny.exe amsterd -mmlog
The Mephisto amsterd is called by default in this configuration without the parameter -nommunlimited, if you have used it without this parameter it is possible that the Mephisto Amsterdam has played with all the force of your computer and not with the original frequency of 12 MHz, that should be checked in the log file it generates.

On your computer this could assume that the Mephisto played with a frequency maybe 32 times the original. And that could mean a lot of points. If we win 35 Elo points every time we double that could be about 175 points, 250 points if the Mephisto wins 50 points when doubling, even 350 points if the Mephisto wins 70 points when doubling.

On the other hand, it is possible that my engine when playing as a human needs more time for the regulation to be better, I think I had discovered it even though I had forgotten it over time, I think it needs at least 15 (30) seconds per game or games 15 (30) minutes, otherwise play loose.

I'm going to play some games against Mephisto Amsterdam with time control st5 and st20, I think in the second control will do much better.
Right I use that default setting and I got this in the log.

Code: Select all

03 Jan 2017 08:18:48 - Mess Mephisto Version: 0.138.d Winboard (Jul  5 2010)

03 Jan 2017 08:18:48 - Mephisto Amsterdam

03 Jan 2017 08:18:48 - Display:  550
03 Jan 2017 08:18:48 - Display: 0441
03 Jan 2017 08:18:48 - Display: d7C7
03 Jan 2017 08:18:48 - Display: F6G5
03 Jan 2017 08:18:48 - Break: time used 3473 > Movetime: 3472
03 Jan 2017 08:18:48 -  ->Time Over
03 Jan 2017 08:18:48 - Display: d7C7
03 Jan 2017 08:18:48 - ENGINE Output: 01  550 0 0 d7c7 f6g5 d7c7
03 Jan 2017 08:18:48 - ENGINE Output: move d7c7
03 Jan 2017 08:18:48 - Display: PLAy
03 Jan 2017 08:18:48 - Emulator org. clock : 12000000
03 Jan 2017 08:18:48 - Emulator curr.clock : 3000000
03 Jan 2017 08:18:48 - Speed factor clock  : 0.25

03 Jan 2017 08:18:48 - OS ticks_per_second : 3312812

03 Jan 2017 08:18:48 - Time in ms for 1 sec: 8
03 Jan 2017 08:18:48 - Speed factor time   : 125.00

03 Jan 2017 08:18:48 - Speed factor total  : 31.25

03 Jan 2017 08:18:48 - ReferenceTimePerSec : 94 (This system has: 8)
03 Jan 2017 08:18:48 - Factor time corr.   : 0.09
03 Jan 2017 08:18:48 - g_tc_delay (ms)     : 170
Later I will use

Code: Select all

-nommunlimited
and run a match with CDrill.
User avatar
pedrox
Posts: 1056
Joined: Fri Mar 10, 2006 6:07 am
Location: Basque Country (Spain)

Re: CDrill - Mephisto Amsterdam Rapid

Post by pedrox »

Do not forget to play with one of the 7 controls that the Mephisto Amsterdam knows, or else it will play with the default level.

Code: Select all

1158 <second: Supported Levels: 
1158 <second: 
1158 <second: Level 0: 3   seconds/move   -> st 3 
1159 <second: Level 1: 5   seconds/move   -> st 5 
1159 <second: Level 2: 10  seconds/move   -> st 10 
1159 <second: Level 3: 20  seconds/move   -> st 20 
1159 <second: Level 4: 60  seconds/move   -> st 60 
1160 <second: Level 5: 120 seconds/move   -> st 120 
1160 <second: Level 6: 40 move in 2 hours -> level 40 120 0
If your configures st5, the machine does not use exactly 5 seconds per move, but it is a mean, some moves can think more than 5 seconds and some less.
Ferdy
Posts: 4848
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: CDrill - Mephisto Amsterdam Rapid

Post by Ferdy »

pedrox wrote:Do not forget to play with one of the 7 controls that the Mephisto Amsterdam knows, or else it will play with the default level.

Code: Select all

1158 <second: Supported Levels: 
1158 <second: 
1158 <second: Level 0: 3   seconds/move   -> st 3 
1159 <second: Level 1: 5   seconds/move   -> st 5 
1159 <second: Level 2: 10  seconds/move   -> st 10 
1159 <second: Level 3: 20  seconds/move   -> st 20 
1159 <second: Level 4: 60  seconds/move   -> st 60 
1160 <second: Level 5: 120 seconds/move   -> st 120 
1160 <second: Level 6: 40 move in 2 hours -> level 40 120 0
If your configures st5, the machine does not use exactly 5 seconds per move, but it is a mean, some moves can think more than 5 seconds and some less.
I use winboard for now at st 3, but there are times MA thinks for more than 5 minutes. So I stop testing.

Do you know if MA will play weaker because of other algorithms, or it only plays weaker because of the time control assuming same hardware speed? Like Level 5 is stronger than Level 1 because of average time.