Rotor 0.8 released

[Jan Brouwer]

Hi and happy pi day,

I have made an update of Rotor with (I hope) a modest strength increase.
The new version can be found at https://sites.google.com/site/chessrotor/

Changes since the previous version are:
- modified transposition table
- tuned mobility evaluation
- changed null move reduction factor R to 3

PS: since I don't have access to an Android device at the moment, I have not been able to test the Android executable.

Share and enjoy,

Jan

[Graham Banks]

Thanks Jan.

[Modern Times]

Does Rotor use tablebases ?

[Modern Times]

Thanks, yes I couldn't see a UCI parameter for it which suggested not.

[Jan Brouwer]

Hi Ray,

As Gabor already mentioned, Rotor does not support table bases.

Jan

[Jan Brouwer]

Hi

The released Win32 executable of Rotor 0.8 still had its affinity mask set to 2 (CPU1).
I use this to make testing a bit reproducible, and had forgotten to change it back before releasing it.
The consequence is that Rotor may use less CPU than expected in certain situations, for example if you run two matches on a multi-core processor with each match Rotor playing against another program.
The processor affinity of an executable can be changed using a program such as imagecfg.

I have uploaded an executable with a corrected affinity mask to https://sites.google.com/site/chessrotor/ (same version, 0.8)
The executable is otherwise identical to the previous one.

Another item to add to my release check-list

Many thanks to Werner Schüle for reporting this issue.

Jan

[geots]

Hi

The released Win32 executable of Rotor 0.8 still had its affinity mask set to 2 (CPU1).
I use this to make testing a bit reproducible, and had forgotten to change it back before releasing it.
The consequence is that Rotor may use less CPU than expected in certain situations, for example if you run two matches on a multi-core processor with each match Rotor playing against another program.
The processor affinity of an executable can be changed using a program such as imagecfg.

I have uploaded an executable with a corrected affinity mask to https://sites.google.com/site/chessrotor/ (same version, 0.8)
The executable is otherwise identical to the previous one.

Another item to add to my release check-list

Many thanks to Werner Schüle for reporting this issue.

Jan

It makes me wonder about Rotor 0.8, and did I miss an exe somewhere. I guess because of the way you refer to the exe with the problem as "the win 32 exe". And your new download refers to it the same way. The new zip is named "rotor_08_win32". Usually when an author refers to it each time as "win 32", he does that to distinguish it from his 64 bit exe so there will be no confusion. I just want to be sure there is no 64bit exe and somehow I missed it.



Best,

george

[Jan Brouwer]

Hi

The released Win32 executable of Rotor 0.8 still had its affinity mask set to 2 (CPU1).
I use this to make testing a bit reproducible, and had forgotten to change it back before releasing it.
The consequence is that Rotor may use less CPU than expected in certain situations, for example if you run two matches on a multi-core processor with each match Rotor playing against another program.
The processor affinity of an executable can be changed using a program such as imagecfg.

I have uploaded an executable with a corrected affinity mask to https://sites.google.com/site/chessrotor/ (same version, 0.8)
The executable is otherwise identical to the previous one.

Another item to add to my release check-list

Many thanks to Werner Schüle for reporting this issue.

Jan

It makes me wonder about Rotor 0.8, and did I miss an exe somewhere. I guess because of the way you refer to the exe with the problem as "the win 32 exe". And your new download refers to it the same way. The new zip is named "rotor_08_win32". Usually when an author refers to it each time as "win 32", he does that to distinguish it from his 64 bit exe so there will be no confusion. I just want to be sure there is no 64bit exe and somehow I missed it.



Best,

george

Hi George,

That is correct, there is no 64 bit executable.
I tested compiling for 64 bit, but there was no difference in speed compared to the 32 bit compile, probably because Rotor still uses the old-fashioned mailbox approach (no bitboards), and its memory foot-print is quite low.

Jan

[geots]

Hi

The released Win32 executable of Rotor 0.8 still had its affinity mask set to 2 (CPU1).
I use this to make testing a bit reproducible, and had forgotten to change it back before releasing it.
The consequence is that Rotor may use less CPU than expected in certain situations, for example if you run two matches on a multi-core processor with each match Rotor playing against another program.
The processor affinity of an executable can be changed using a program such as imagecfg.

I have uploaded an executable with a corrected affinity mask to https://sites.google.com/site/chessrotor/ (same version, 0.8)
The executable is otherwise identical to the previous one.

Another item to add to my release check-list

Many thanks to Werner Schüle for reporting this issue.

Jan

It makes me wonder about Rotor 0.8, and did I miss an exe somewhere. I guess because of the way you refer to the exe with the problem as "the win 32 exe". And your new download refers to it the same way. The new zip is named "rotor_08_win32". Usually when an author refers to it each time as "win 32", he does that to distinguish it from his 64 bit exe so there will be no confusion. I just want to be sure there is no 64bit exe and somehow I missed it.



Best,

george

Hi George,

That is correct, there is no 64 bit executable.
I tested compiling for 64 bit, but there was no difference in speed compared to the 32 bit compile, probably because Rotor still uses the old-fashioned mailbox approach (no bitboards), and its memory foot-print is quite low.

Jan

Thank you very much for taking the time to stop and help me.




Best,

george

[jshriver]

Very cool, I look forward to testing it!

Have a good weekend.
-Josh
http://olympuschess.com/tourney