RobboLito 2 is available !

i will the link of this engine

Vladimir wrote:

+38 ELO. Nice, Man!

kranium Returns (2012)

thx Vladimir

still have a long way to go to catch Houdini however

rufy-ace wrote:

i will the link of this engine

i will release it as soon as this final test is over

Wins = 2158
Draws = 4646
Losses = 1278
Av.Op. Elo = 3200

Result : 4481.0/8082 (+2158,=4646,-1278)
Perf. : 55.4 %
Margins :
68 % : (+ 0.4,- 0.4 %) -> [ 55.1, 55.8 %]
95 % : (+ 0.7,- 0.7 %) -> [ 54.7, 56.1 %]
99.7 % : (+ 1.1,- 1.1 %) -> [ 54.4, 56.5 %]

Elo : 3238
Margins :
68 % : (+ 3,- 2) -> [3235,3240]
95 % : (+ 5,- 5) -> [3233,3243]
99.7 % : (+ 7,- 7) -> [3231,3245]

games:
mediafire.com download.php?8e2pln4l3012tlt

engine:
http://www.chesslogik.com/

Last edited by kranium on Sun Sep 30, 2012 8:52 am; edited 1 time in total

kranium wrote:

rufy-ace wrote:

i will the link of this engine

i will release it as soon as this final test is over

And if he's about ready? how many games need to be played?

kranium wrote:

games:
mediafire.com download.php?8e2pln4l3012tlt

Elostat (Start Elo = 3200)

Chess100 wrote:

Quocvuong82 wrote:

Hi all-

the test is looking really good...
if it continues well, I'll release RobboLito 0.20Q x64 SSE4.2 later today

Code:

Wins  = 315
Draws  = 635
Losses = 171
Av.Op. Elo = 3200

Result    : 632.5/1121 (+315,=635,-171)
Perf.      : 56.4 %
Margins    :
 68 %      : (+  1.0,-  1.0 %) -> [ 55.5, 57.4 %]
 95 %      : (+  1.9,-  1.9 %) -> [ 54.5, 58.3 %]
 99.7 %    : (+  2.9,-  2.9 %) -> [ 53.5, 59.3 %]

Elo        : 3245
Margins    :
 68 %      : (+  7,-  7) -> [3238,3252]
 95 %      : (+ 13,- 13) -> [3232,3258]
 99.7 %    : (+ 20,- 20) -> [3225,3265]

Hello Quocvuong82,

What utility you use to calculate this?

Greetings.

I used EloStat

RobboLito 0.20Q x64 SSE4.2
http://www.chesslogik.com/downloads/RobboLito_020Q.rar

- Much evaluation code added: pins, queen, rook, pawn, and endgame
- Q optimizations (Quocvuong82)
- Agressive hash prefetching
- Windows large page support
- 64 bit Intel intrinsics
- 10% speed increase

Mirror: RobboLito 0.20Q

kranium wrote:

engine:
http://www.chesslogik.com/

What about RobboLito 0.20Q x64 (non-SSE4.2) ?

kranium wrote:

Code:

Games Completed = 8082 of 50000 (Avg game length = 13.094 sec)
Settings = RR/32MB/1000ms+100ms/M 1000cp for 12 moves, D 150 moves/EPD:openings.epd(4000)
Time = 18802 sec elapsed, 97517 sec remaining
 1.  IvanHoe 9.46b x64           3601.0/8082   1278-2158-4646     (L: m=1097 t=0 i=0 a=1061)   (D: r=3321 i=703 f=384 s=10 a=228)   (tpm=98.1 d=12.95 nps=2876399)
 2.  RobboLito 0.20Q x64 SSE4.2   4481.0/8082   2158-1278-4646     (L: m=898 t=0 i=0 a=380)   (D: r=3321 i=703 f=384 s=10 a=228)   (tpm=96.2 d=13.28 nps=3554236)

Wins = 2158
Draws = 4646
Losses = 1278
Av.Op. Elo = 3200

Result : 4481.0/8082 (+2158,=4646,-1278)
Perf. : 55.4 %
Margins :
68 % : (+ 0.4,- 0.4 %) -> [ 55.1, 55.8 %]
95 % : (+ 0.7,- 0.7 %) -> [ 54.7, 56.1 %]
99.7 % : (+ 1.1,- 1.1 %) -> [ 54.4, 56.5 %]

Elo : 3238
Margins :
68 % : (+ 3,- 2) -> [3235,3240]
95 % : (+ 5,- 5) -> [3233,3243]
99.7 % : (+ 7,- 7) -> [3231,3245]

games:
mediafire.com download.php?8e2pln4l3012tlt

engine:
http://www.chesslogik.com/

Thanks kranium.

Bob wrote:

kranium wrote:

engine:
http://www.chesslogik.com/

What about RobboLito 0.20Q x64 (non-SSE4.2) ?

this engine is geared toward (and has a lot of code for) modern intel family processors...

i intend to make a different compile but it takes time...
please be patient...
(dont forget i'm producing this engine in my free time)

Thanks a lot from me too

Keep up the good work, Kranium

Farbror

farbrorsniff wrote:

Thanks a lot from me too
Keep up the good work, Kranium
Farbror

great... for you kranium... nice work....

TC : 10 min + 5 sec

RobboLito 0.20Q SSE4.2 vs. Houdini 2c Pro?

Augusto_19 wrote:

RobboLito 0.20Q SSE4.2 vs. Houdini 2c Pro?

vs Critter 1.6a
https://www.chess2u.com/t5277p15-robbolito-tests#37469

Hi all-

in addition to RobboLito 0.20Q x64 SSE4.2.exe,
2 new compiles are available:

RobboLito 0.20Q w32.exe
RobboLito 0.20Q x64.exe

www.chesslogik.com

RobboLito 0.20Q w32 says msvcr100.dll is missing.

msvcr100.dll : mediafire.com ?tgmwkwdacjwozmy

Steve.R wrote:

RobboLito 0.20Q w32 says msvcr100.dll is missing.

msvcr100.dll : mediafire.com ?tgmwkwdacjwozmy

thanks Steve

i've recompiled it to no longer need a Microsoft run-time DLL
if you prefer such an executable, please download the package again
(speeds are the same)

kranium, I might hold an IPPOLIT VS THE WORLD match (If Ippolit is the correct term, I don't know)

Do you think the 10 best engines in the world can beat the 10 best ippolit-based engines?

Castle00 wrote:

kranium, I might hold an IPPOLIT VS THE WORLD match (If Ippolit is the correct term, I don't know)

Do you think the 10 best engines in the world can beat the 10 best ippolit-based engines?

no

"The open-source development of Fire has officially ended, and it is no longer available for download."

Hmmm, what is the difference between Fire and RobboLito2?

waelsae wrote:

"The open-source development of Fire has officially ended, and it is no longer available for download."

Hmmm, what is the difference between Fire and RobboLito2?

Good question...
well, for one, Fire had many features that Robbo does not:
multi positional gain (for each CPU/core/thread)
multi history (for each CPU/core/thread)
old and new robbobase support (up to 6 piece endings)
70+ UCI/fire.cfg options
random configuration generation
optional timestamped I/O log file
and more...
it's fatter and slower

Robbo is much faster, highly optimized for speed, and is considerably stronger (probably +30 ELO or more, test it yourself)
it does more agressive pruning
material values, search and evaluations are are also quite different:
for ex: here's a random position:
http://www.chesslogik.com/Robbo_vs_Fire.jpg

there are also similarities, of course, having both been developed from Ippolit source code,
but they lie under the surface, and are pretty hard to see, IMO
i.e. basic functions like hash tables, SMP algorithms, move generation, etc.
take a look at the source code if you want to know more