Winboard Forum

PikoSzachy is a new polish engine for WinBoard.

The name pikoSzachy (in english: picoChess) refers to the very small size of the program - only 10 KB!
It's a simplified version of my another engine - nanoSzachy.
PikoSzachy can use opening book of nanoSzachy.

Download:
pikoSzachy:
http://www.kalisz.mm.pl/~pic/nanochess/index.html#picoChess
nanoSzachy:
http://www.kalisz.mm.pl/~pic/nanochess/

Piotr Cichy wrote:PikoSzachy is a new polish engine for WinBoard.

The name pikoSzachy (in english: picoChess) refers to the very small size of the program - only 10 KB!

Hi Piotr,

This is truly astonishing. I thought my engine Glaurung, with an executable size of around 70 KB was rather small. Any hints about how you achieve such tiny executable sizes? How big is the source code?

Tord

Tord Romstad wrote:Hi Piotr,

This is truly astonishing. I thought my engine Glaurung, with an executable size of around 70 KB was rather small. Any hints about how you achieve such tiny executable sizes? How big is the source code?

Tord

Hi Tord.

Main reasons of small program size are:
- very compact source code (about 30KB, written in C++)
- code optimized for size, not speed
- using pure C++, avoiding external static linked libraries
- most big tables are precomputed at start, not kept in program
- small number of universal functions with many parameters
(for example, normal and quiescent search use common code)
- avoiding identical or similiar code in different parts of code

Piotr Cichy wrote:PikoSzachy is a new polish engine for WinBoard.

The name pikoSzachy (in english: picoChess) refers to the very small size of the program - only 10 KB!
It's a simplified version of my another engine - nanoSzachy.
PikoSzachy can use opening book of nanoSzachy.

Download:
pikoSzachy:
http://www.kalisz.mm.pl/~pic/nanochess/index.html#picoChess
nanoSzachy:
http://www.kalisz.mm.pl/~pic/nanochess/

Piotr,
How does the strength of PikoSzachy compare to nanoSzachy? nanoszachy is one of my favorite engines to test against.

eric_oldre wrote:Piotr,
How does the strength of PikoSzachy compare to nanoSzachy? nanoszachy is one of my favorite engines to test against.

I compiled pikoSzachy last Sunday and did not have too much time for testing. But in my first tests (mainly short time controls) the strength of playing seems to be only a litlle worse. I hope to see more results soon in test tourneys.

PikoSzachy has different move generator, does not use bitboards and is about 25% slower. Evaluation function is similiar in both programs. I tried also a simpler (and shorter) evaluation function, but playing strength was noticably worse (in spite of achieving about 2 times more nodes per second).

Jim.

Jim Ablett wrote:

Hi Jim.

Thanks for the new logo