Page 1 of 4
FruitFly 1.1
Posted: Fri Jul 20, 2012 10:29 am
by ZirconiumX
I won't even try to put a description of the changes, because I strongly doubt that anyone will read this.
Download.
Matthew:out
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 10:52 am
by kgburcham
ZirconiumX wrote:I won't even try to put a description of the changes, because I strongly doubt that anyone will read this.
Download.
Matthew:out
I read most all of your posts Matthew.
kgburcham
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 10:59 am
by Vinvin
ZirconiumX wrote:I won't even try to put a description of the changes, because I strongly doubt that anyone will read this.
Download.
Matthew:out
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 11:04 am
by ZirconiumX
Ok, I have a response. This is a plus. I'd rate it (being pessimistic) still slightly weaker than Fruit 2.1, but with a completely different playing style.
I'd say that FruitFly 1.1 is arrogant, to an extent.
Main changes:
Complete rewrite which is what has taken me so long.
Eval:
New material values; untuned.
King tropism; untuned.
Tempo bonus; untuned.
Superlinear interpolator; untuned.
Blocked fianchetto penalty; untuned.
Search:
Quite a bit taken (externally, I have not RE'd) Fruit 05/11/03; untuned by me.
Sliding LMR as used by Rodent; untuned.
Futility done at depth <= 3; untuned.
Aspirated search; thanks Harald; untuned.
Static nullmove; untuned.
Double nullmove replaces verification search; untuned.
Late move Pruning; untuned.
If you need to know what any of the settings mean, just ask.
Matthew:out
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 11:05 am
by velmarin
Sometimes we read, but nothing is said ...
And download.
Indeed there is only code, not executables.
Look in your github yesterday 4 months and nothing changes.
Encouragement to work, the results arrived.
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 11:08 am
by ZirconiumX
Dad wiped my hard drive. Nuff said.
Is anyone at all impressed at this for saying I am just a 12 year old?
Also, as I have only an apple computer, and an old one at that, I haven't been able to produce any executables.
Matthew:out
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 11:14 am
by velmarin
Now I have little time,
wanted to do a quick build,
but it is not possible at this time.
Code: Select all
***** ClCompile (X64 - Intel C++)
1> attack.cpp
1> board.cpp
1> book.cpp
1> eval.cpp
1>eval.cpp(251): error : more than one instance of overloaded function "abs" matches the argument list:
1> function "abs(int)"
1> function "abs(long)"
1> function "abs(__int64)"
1> argument types are: (unsigned int)
1> tropism = 14 - (abs(COL(from) - COL(dest)) + abs(ROW(from) - ROW(dest)));
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 11:18 am
by ZirconiumX
I *know* the file extension says c++, but can you try using the Intel C compiler?
It compiles perfectly using GCC on my computer.
Matthew:out
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 11:25 am
by ZirconiumX
This is obviously a problem with your c++ library, as
abs should only have int and long.
I have uploaded a new version which uses a preprocessor macro ABS() instead of the C standard library abs() in eval.cpp.
Download it again, and report back.
Matthew:out
Re: FruitFly 1.1
Posted: Fri Jul 20, 2012 11:37 am
by velmarin
I also happened to Microsoft compiler,
I know it's a minor problem.
But now I must get away from the computer.
If no after compilation.
What I compilare. (in two hours)