would she say a bigger program is a better program?

[kgburcham]

Are these correct?
Is it correct to say these values are the size of only the engine code?
When we use the words open source, are these the values that would measure the size of the open source, excluding the interface of course?
I must be missing something, surely there is not this much variation.

Note how Rybka size doubled between versions 1.0 and 2.3.2a.
Note how Rybka is much larger than other programs.
Note how small Deep Sjeng is.

Rybka 2.3.2a mp 64 bit (4.12mb)
Rybka 2.3.2a mp 32 bit (3.64mb)
Rybka 1.0 Beta 64 bit (2.40mb)

Crafty 22.1 (1.73mb)

Deep Shredder 11 UCI 64 bit (1.48mb)

Zappa Mexico II 64 bit (1.38mb

Glaurung 2.01 64 bit (1.18mb)

movei00 8 438 (500kb)

Little Thought mp 64 bit (373kb)

Deep Sjeng 3.0 32 bit (264kb)
Deep Sjeng 3.0 64 bit (264kb)

nanoSzachy (29.5 KB)
pikoSzachy (9 KB)

[Tony Thomas]

In my opinion the size of the executable has a lot to do with the compiler optimizations being used. We know that Rybka is large because it contains those data tables..I based my opinion on the fact that the size of Romichess almost doubles when I compare the versions from Michael and Jim. I doubt that bigger is better, but who am I to say that when the biggest is currently the strongest??

[Uri Blass]

Are these correct?
Is it correct to say these values are the size of only the engine code?
When we use the words open source, are these the values that would measure the size of the open source, excluding the interface of course?
I must be missing something, surely there is not this much variation.

Note how Rybka size doubled between versions 1.0 and 2.3.2a.
Note how Rybka is much larger than other programs.
Note how small Deep Sjeng is.

Rybka 2.3.2a mp 64 bit (4.12mb)
Rybka 2.3.2a mp 32 bit (3.64mb)
Rybka 1.0 Beta 64 bit (2.40mb)

Crafty 22.1 (1.73mb)

Deep Shredder 11 UCI 64 bit (1.48mb)

Zappa Mexico II 64 bit (1.38mb

Glaurung 2.01 64 bit (1.18mb)

movei00 8 438 (500kb)

Little Thought mp 64 bit (373kb)

Deep Sjeng 3.0 32 bit (264kb)
Deep Sjeng 3.0 64 bit (264kb)

nanoSzachy (29.5 KB)
pikoSzachy (9 KB)

Size has nothing to do with playing strength.

Strelka is clearly smaller than rybka beta and based on rybka beta.

Size of movei is result of some functions that are not related to playing strength and I could make it smaller with the same playing strength.

Note that probing the nalimov tablebases can increase the size without significant effect on playing strength.

Uri

[Jouni]

But note that Rybka 3 is again smaller = "only" 2 580 480 bytes! If Vasik wants to make self-extracting packed exe-file it would be a lot smaller (around 850 000 bytes).

Jouni

[Jouni]

Toga II 1.4 is also stunningly small with 225 280 bytes only! And it gives the best score against Rybka among my engines...

Jouni