Home made engine ranking: Critter 1.6a??

[Chessbykermy]

Dear all,

As a correspondence chess player, I have several engines to test critical positions. So, some days ago I decided to put them all into a little test:
1. 34 positions.
2. Computer: my old laptop: Macbook pro 10 years old.
3. MPV = 1.
4. Time for position: 7 minutes.

The standings (as % of test solved):

1. AsmFish...............81.7%
2. Critter 1.6a..........80.0%
3. Komodo 12.1.........70.0%
4. McBrain 9.1..........68.3%
5. Stockfish 10..........68.3%
6. Sting SF11............58.3%
7. HIARCS 14............51.7%

I'm surprised by Critter results, this old guy has finished in second position ... really close to AsmFish. In my opinion, Critter seems to be very good solving tactical positions, where a piece sacrifice is needed.

I would appreciate your opinion about Critter. Do you think is a good engine for solving tests nowadays?? or the big ones (Komodo, Stockfish, ...) should be the only opinion to take into account?

[ernest]

... and I am surprised at your results, with asmFish being so much ahead of Stockfish 10 !

Maybe it's an error-bar problem...

[Dann Corbit]

... and I am surprised at your results, with asmFish being so much ahead of Stockfish 10 !

Maybe it's an error-bar problem...

WIth 37 positions, it is easy to get a big percentage change by a few less solutions.

[Jouni]

Critter 1.6a is basically Houdini 1.5. And Houdini 1.5 is fastest in some testsuites. Doen't mean ANYTHING .

[Guenther]

Dear all,

As a correspondence chess player, I have several engines to test critical positions. So, some days ago I decided to put them all into a little test:
1. 34 positions.
2. Computer: my old laptop: Macbook pro 10 years old.
3. MPV = 1.
4. Time for position: 7 minutes.

The standings (as % of test solved):

1. AsmFish...............81.7%
2. Critter 1.6a..........80.0%
3. Komodo 12.1.........70.0%
4. McBrain 9.1..........68.3%
5. Stockfish 10..........68.3%
6. Sting SF11............58.3%
7. HIARCS 14............51.7%

I'm surprised by Critter results, this old guy has finished in second position ... really close to AsmFish. In my opinion, Critter seems to be very good solving tactical positions, where a piece sacrifice is needed.

I would appreciate your opinion about Critter. Do you think is a good engine for solving tests nowadays?? or the big ones (Komodo, Stockfish, ...) should be the only opinion to take into account?

Have you changed the default settings of Critter?
People always forget that Critter uses 4 threads by default...

Moreover a few test positions, no matter how nicely selected, will never give an appropriate picture.

[mar]

Critter 1.6a is basically Houdini 1.5. And Houdini 1.5 is fastest in some testsuites. Doen't mean ANYTHING .

Where did you come up with that nonsense? Critter has a long history in rating lists, long before H was "born".
First Houdini was basically 1:1 clone of "Robbolito", which was ... well
Where Houdart steals from today, I don't know. But I have an idea

[corres]

Have you changed the default settings of Critter?
People always forget that Critter uses 4 threads by default...

??
Sometimes I also use Critter 1.6a (64 bits) without issues.
Critter sets the number of threads automatically, as I have experienced.
I use PC of eight cores.

[Guenther]

Have you changed the default settings of Critter?
People always forget that Critter uses 4 threads by default...

??
Sometimes I also use Critter 1.6a (64 bits) without issues.
Critter sets the number of threads automatically, as I have experienced.
I use PC of eight cores.

Shouldn't you be able to read uci options yourself?

Code: Select all

Critter v1.6a 64-bit, by Richard Vida
hardware POPCNT disabled
uci
id name Critter 1.6a 64-bit
id author Richard Vida
option name Hash type spin default 64 min 8 max 8192
option name Ponder type check default false
option name OwnBook type check default true
option name Book File type string default book.cbk
option name UCI_Chess960 type check default false
option name MultiPV type spin default 1 min 1 max 100
--------------------------------------------------------------------
option name Threads type spin default 4 min 1 max 32
--------------------------------------------------------------------
option name Split Depth type spin default 5 min 4 max 15
option name GaviotaTbPath type string default /gtb
option name GaviotaTbCache type spin default 32 min 4 max 256
option name GaviotaTbCompression type combo default cp4 var uncompressed var cp1
 var cp2 var cp3 var cp4
option name Tablebase Usage type combo default Only At Root var Disable var Only
 At Root var Everywhere
option name Clear Hash type button
option name Use Session File type check default false
option name Session File type string default session.csf
option name SF Size type spin default 32 min 2 max 1024
option name SF Write Depth type spin default 8 min 2 max 32
option name SF Material Limit type spin default 0 min 0 max 32
option name SF Move Limit type spin default 0 min 0 max 999
option name Clear SF type button
option name Resolve Score Drops type check default false
option name King Safety Weight type spin default 32 min 0 max 100
uciok

[Chessbykermy]

I thought Threads should be related with number of cores: 1 cores = 1 thread, 2 cores = 2 threads, 4 cores = 4 threads, ...

So I use this rule to modify engine settings. It's that rule incorrect?

Thanks!

[Guenther]

I thought Threads should be related with number of cores: 1 cores = 1 thread, 2 cores = 2 threads, 4 cores = 4 threads, ...

So I use this rule to modify engine settings. It's that rule incorrect?

Thanks!

If you had set 1 thread explicitely in your GUI or cmd all should be ok.