optimize 1366x768

John Stanback (home)
John Stanback (CPW)
Zarkov (CPW)

 



UCI Engine Wasp
by John Stanback,
... current version: 2.01
back to Frank's Chess Page

® Copyright: John Stanback
Last changes: June 14th, 2017, 09:30

 

June 16th, 2016 (start point): In the past John sent me often versions of Zarkov. I tested unprofessional his engine, produced a lot of games but not more (Winboard versions of Zarkov are in the past not official available). John told me that he is working on a new engine, called Wasp. Reason enough to do a bit for him, I hope more professional as in the past. I will try to start here a beta test selection. If you like you can follow the work between John and myself. But most important for me is that the program is available for all (I tested available engines only) I hope my english is good enough for that what I will do here. Messages by John in green ...

 

 

May 28th, June 05th, June 14th, 2017:

Now 4.000 games vs. the same group of opponents with:

1. Wasp 20170527 BMI2 x64 (dev. version, around 15-20 Elo stronger as Wasp 2.00 / 2.01)
2. FEOBOS v4.0 -mt -draw (Wasp should be play no bad lines, and a very high book contempt will the result).

Test-run 001: 2.835 Elo after 1.000 games = 47,80%, 478.0 - 522.0 points
Test-run 002: 2.833 Elo after 1.000 games = 47,45%, 474.5 - 525.5 points
Test-run 003: 0.000 Elo after 1.000 games = 00,00%, 000.0 - 000.0 points
Test-run 004: 0.000 Elo after 1.000 games = 00,00%, 000.0 - 000.0 points  = 0.000.0 - 0.000.0 points = 00,00%

FEOBOS v4.0 / Wasp 20170527 BMI2 x64 / after Test-run 002
Games : 2000 (finished)

White Wins :   580 (29.0 %)
Black Wins :   447 (22.4 %)
Draws :   973 (48.6 %)
Unfinished : 0

White Perf. : 53.3 %
Black Perf. : 46.7 %

ECO A =   284 Games (14.2 %)
ECO B =   526 Games (26.3 %)
ECO C =   324 Games (16.2 %)
ECO D =   354 Games (17.7 %)
ECO E =   512 Games (25.6 %)

  

 

May 01st / May 09th / May 18th / May 26th, 2017:

Wasp 2.00 BMI2 x64 is playing 4 times the same 1.000 games test-run. I am using for the moment a prototyp version of my new opening book. This prototyp version produced not a perfect assignment to the A00-E99 ECO codes (rarely ECO codes should be play rarely and popular ECO codes should be play often) . Much rarely ECO codes are playing to often, much other ECO codes missed completely. I am using for the Wasp test-runs 20 of my favorite still under development engines, in the near of the ELO performance of Wasp. OK, much older versions with a nice playing style are included too (Spark, Junior, Hiarcs) and of course different clearly stronger engines should be included too (Chiron, Booot).

Note: The test-runs will be run very slowly (around 120 games per day). I am using one i7-4770k system only! 4 games are still running to the same time only!

Test-run 001: 2.805 Elo after 1.000 games = 43,50%, 435.0 - 565.0 points
Test-run 002: 2.820 Elo after 1.000 games = 45,65%, 456.5 - 543.5 points
Test-run 003: 2.814 Elo after 1.000 games = 44,80%, 448.0 - 552.0 points
Test-run 004: 2.820 Elo after 1.000 games = 45,65%, 456.5 - 543.5 points  = 1.796.0 - 2.204.0 points = 44,90%

 

Wasp 2.00 BMI2 x64 - Chiron 4 x64                47.5 - 152.5  +11/= 73/-116  23.75%
Wasp 2.00 BMI2 x64 - Booot 6.1 POP x64           48.0 - 152.0  + 5/= 86/-109  24.00%
Wasp 2.00 BMI2 x64 - Texel 1.07a9 x64            61.5 - 138.5  +22/= 79/- 99  30.75%
Wasp 2.00 BMI2 x64 - iCE 3.0 v658 POP x64        66.5 - 133.5  +26/= 81/- 93  33.25%
Wasp 2.00 BMI2 x64 - SmarThink 1.97 SSE4 x64     67.0 - 133.0  +26/= 82/- 92  33.50%
Wasp 2.00 BMI2 x64 - Nemorino 2.00a POP x64      88.5 - 111.5  +32/=113/- 55  44.25%
Wasp 2.00 BMI2 x64 - Hakkapeliitta TCEC v2 x64   86.0 - 114.0  +37/= 98/- 65  43.00%
Wasp 2.00 BMI2 x64 - Pedone 1.5 BMI2 x64         87.5 - 112.5  +33/=109/- 58  43.75%
Wasp 2.00 BMI2 x64 - Arasan 20.0 POP x64         89.5 - 110.5  +40/= 99/- 61  44.75%
Wasp 2.00 BMI2 x64 - Vajolet2 2.3 POP x64        93.0 - 107.0  +34/=118/- 48  46.50%
Wasp 2.00 BMI2 x64 - Hiarcs 14 WCSC w32          89.0 - 111.0  +41/= 96/- 63  44.50%
Wasp 2.00 BMI2 x64 - Bobcat 8.0 x64              87.5 - 112.5  +37/=101/- 62  43.75%
Wasp 2.00 BMI2 x64 - Laser 1.3 x64              104.5 -  95.5  +52/=105/- 43  52.25%
Wasp 2.00 BMI2 x64 - Cheng 4.39 x64             102.5 -  97.5  +49/=107/- 44  51.25%
Wasp 2.00 BMI2 x64 - Crafty 25.2 x64            102.5 -  97.5  +61/= 83/- 56  51.25%
Wasp 2.00 BMI2 x64 - Junior 13.3.00 x64         108.5 -  91.5  +64/= 89/- 47  54.25%
Wasp 2.00 BMI2 x64 - Rodent III 0.172 x64       110.0 -  90.0  +64/= 92/- 44  55.00%
Wasp 2.00 BMI2 x64 - Spark 1.0 x64              116.5 -  83.5  +59/=115/- 26  58.25%
Wasp 2.00 BMI2 x64 - Gaviota 1.0 AVX x64        116.5 -  83.5  +70/= 93/- 37  58.25%
Wasp 2.00 BMI2 x64 - Amoeba 2.4.x64             122.5 -  77.5  +77/= 91/- 32  61.25%

55 games lost undo move number 60, 113 games won = 67,26% vs. average of light stronger group of opponents!
10 games lost undo move number 50,  37 games won = 78,70% vs. average of light stronger group of opponents!

 

Wasp is very strong in the first gaming phase, typical Wasp statistics!
One of the reasons that Wasp is one of my engine favorits! 

Since game number 2.100 / 4.000 I am using FEOBOS v3.0 without move transpositions. D00-D99 could be ~5% higher and C00-C99 could be ~5% lesser. But I like that a lot. 5% more E00-E99 = much more interesting positions for computerchess matches in my humble opinion. My dream opening book was really born with FEOBOS. Most computer chess opening books have not enough to E00-E99. In the last years I am searching _more_ for E00-E99 balanced lines for my FCP Live Book and now the results can be see. Yes, my FEOBOS book is over optimated to E00-E99 opening systems and this is right!

 

FEOBOS v3.0 / Wasp 2.00 BMI2 x64
Games : 1900 (finished)

White Wins :   588 (30.9 %)
Black Wins :   417 (21.9 %)
Draws :   895 (47.1 %)
Unfinished : 0

White Perf. : 54.5 %
Black Perf. : 45.5 %

ECO A =   277 Games (14.6 %)
ECO B =   495 Games (26.1 %)
ECO C =   313 Games (16.5 %)
ECO D =   352 Games (18.5 %)
ECO E =   463 Games (24.4 %)

 

SUPER GM database (different sources), 1990 - end of the year 2015
Games : 135430 (finished)

White Wins : 47252 (34.9 %)
Black Wins : 29841 (22.0 %)
Draws : 58337 (43.1 %)
Unfinished : 0

White Perf. : 56.4 %
Black Perf. : 43.6 %

ECO A = 18877 Games (13.9 %)
ECO B = 36313 Games (26.8 %)
ECO C = 23767 Games (17.5 %)
ECO D = 30737 Games (22.7 %)
ECO E = 25736 Games (19.0 %)

 

Chessbase Big Database 2017, 2000 - 2017
Games : 147522 (finished)

White Wins : 48536 (32.9 %)
Black Wins : 30823 (20.9 %)
Draws : 68163 (46.2 %)
Unfinished : 0

White Perf. : 56.0 %
Black Perf. : 44.0 %

ECO A = 23587 Games (16.0 %)
ECO B = 37289 Games (25.3 %)
ECO C = 26905 Games (18.2 %)
ECO D = 33741 Games (22.9 %)
ECO E = 25910 Games (17.6 %)

 

 

May 04th, 2017:

Wasp 2.01 released!

John wrote:
Werner Schüle found that Wasp 2.00 was occasionally crashing with ponder ON when a mate was reached during a ponder search. There is a bug in version 2.00 which some GUI's might have been able to deal with. I found and fixed the problem for the attached version 2.01. Change the name to Wasp_201.7z and use 7-zip to extract. Please put this version on your web site.

Here are the update information from John's documentation.

Wasp 2.01: May 3, 2017
- fixed bug when pondering and search times out (usually due to mate score)
- fixed typo in blocked passed pawn penalty

Wasp 2.00: April 18, 2017
- added multi-threading using very lazy SMP approach where each thread simply does an iterative deepening loop on the root position and the only communication between threads is the shared main hash table. as others have found, this works surprisingly well.
- modified futility pruning, late-move pruning, and late-move reduction formulas to increase selectivity and reduce effective branching factor slightly
- simplified extensions-- extend safe checks, safe pawn to 7th, and capture of last enemy piece by one ply.
- tweaked many evaluation parameters
- move generator now has 3 stages where moves are generated & searched: 1. hash move, 2. captures/promotions, 3. quiet moves

If you like to test Wasp please download his engine under: wasp_201.zip (0.651Mb)

Included is:
1. Wasp_201-bmi2-x64.exe (with bit manipulation instructions = BMI2)
2. Wasp_201-x64.exe (with no bit manipulation instructions for older CPU's)
3. Wasp_201-w32.exe (compiled for athlon-64, should also work with Pentium 3 or greater Intel processors)
4. release-notes.txt
5. The two main logos by Wilhelm Hudetz

Thanks John!

 

 

April 21st, 2017:

Wasp 2.00 released!

John wrote:
- Wasp 2.00 can now use multiple threads using "lazy SMP" search.

My testing shows that Wasp 2.00 on a single thread is about 50 Elo better than Wasp 1.25.
At very fast time controls, using 4 threads gains about 130 Elo. 

If you like to test Wasp please download his engine under: wasp_200.zip (0.650Mb)

Included is:
1. Wasp_200-bmi2-x64.exe (with bit manipulation instructions = BMI2)
2. Wasp_200-x64.exe (with no bit manipulation instructions for older CPU's)
3. Wasp_200-w32.exe (compiled for athlon-64, should also work with Pentium 3 or greater Intel processors)
4. release-notes.txt
5. The two main logos by Wilhelm Hudetz

Thanks John! 

 

 

October 08th, 2016:

Wasp 1.25 BMI2 x64 test-run ... the final results!

Wasp 1.25 BMI2 x64 is clearly improved, most in late middlegame.
This was a great update John!

C O M P A R S I O N


Final results
-------------

   # Player             :     Elo  Games  Score%   won  draw  lost  Points  Draw%  Error
  --------------------------------------------------------------------------------------
  30 Wasp 1.25 BMI2 x64 : 2794.63   2450    42.7   499  1096   855  1047.0   44.7   9.83  +43.08 Elo
  37 Wasp 1.01 BMI2 x64 : 2751.55   2450    37.6   371  1101   978   921.5   44.9  10.32



Move number 020-059
-------------------
  23 Wasp 1.25 BMI2 x64 : 2841.69    351    51.9    51   262    38   182.0   74.6  19.19  +14.15 Elo
  30 Wasp 1.01 BMI2 x64 : 2827.54    352    47.9    44   249    59   168.5   70.7  18.61


Move number 060-079
-------------------
  27 Wasp 1.25 BMI2 x64 : 2769.90    738    40.7   194   212   332   300.0   28.7  22.84  +60.22 Elo
  38 Wasp 1.01 BMI2 x64 : 2709.68    784    33.2   145   230   409   260.0   29.3  23.63


Move number 080-099
-------------------
  34 Wasp 1.25 BMI2 x64 : 2741.75    647    37.3   160   163   324   241.5   25.2  23.52  +55.01 Elo
  41 Wasp 1.01 BMI2 x64 : 2686.74    618    30.4   117   142   359   188.0   23.0  25.37


Move number 100-299
-------------------
  33 Wasp 1.25 BMI2 x64 : 2805.22    714    45.3    94   459   161   323.5   64.3  13.61  + 4.16 Elo
  36 Wasp 1.01 BMI2 x64 : 2801.06    696    43.8    65   480   151   305.0   69.0  14.01


  35. Wasp 1.01 x64  :  2785.58    702    44.5    67   491   144   312.5   69.9  14.03  + 45,05 Elo!

 

September 30th, 2016:

Wasp 1.25 released!

John wrote:
- compiled using new version of gcc (6.1.0)
- generate and try hash move (if any) before generating all other moves
- slight change to history table updating
- slight change to pruning in quiescence search
- allow late move pruning and reduction for unsafe checks
- changed late move pruning:
   - prune non-tactical moves after 1+3*depth moves have been tried
   - prune if moves_searched >= 3 and eval < alpha-100*depth (futility pruning)
- changed late move reduction to reduce by 3 plies if depth >= 6 and moves_searched >= 40 - 2*depth
- minor tweaks to most evaluation parameters such as piece values, mobility, king safety, passed pawns ...
- added term for king mobility in endgame
- removed special KRPKR evaluation function
- modified draw adjustment calculation

If you like to test Wasp please download his engine under: wasp_125.zip (0.572Mb)

Included is:
1. Wasp_125-bmi2-x64.exe (with bit manipulation instructions = BMI2)
2. Wasp_125-x64.exe (with no bit manipulation instructions for older CPU's)
3. Wasp_125-w32.exe (compiled for athlon-64, should also work with Pentium 3 or greater Intel processors)
4. release-notes.txt
5. The two main logos by Wilhelm Hudetz

Thanks John!

 

July 02nd, 2016:

Wasp 1.02, bugfix released!

John wrote:
I fixed the problem with sudden-death time controls not working with Fritz GUI.  I also compiled the 32 bit exe such that it should work with older CPU's than the previous version did.  There is no change to search or eval, it should play the same as the previous version.

If you like to test Wasp please download his engine under: wasp_102.zip (0.587Mb)

Included is:
1. Wasp_102-bmi2-x64.exe (with bit manipulation instructions = BMI2)
2. Wasp_102-x64.exe (with no bit manipulation instructions for older CPU's)
3. Wasp_102-w32.exe (compiled for athlon-64, should also work with Pentium 3 or greater Intel processors)
4. release-notes.txt
5. The two main logos by Wilhelm Hudetz

Thanks John & Ferdinand Mosca for the logfiles (TalkChess)!

 

July 01st, 2016:

Chess Programming WIKI updated!

Gerd Isenberg updated the detail information to John Stanback.
John Stanback (CPW)

Thanks Gerd for the work you do for the community ... again and again!

... my symbolic long time Award for Gerd!

 

June 23rd, 2016:

Wasp 1.01, John sent two new executables!

John wrote:
Attached are new Wasp 1.01 executables for 32-bit and 64-bit with no bit manipulation instructions for older CPU's.  I updated the 32 bit exe to display the correct version number (1.01) instead of 1.00.  There is no other change as compared to yesterday's version.

If you like to test Wasp please download his engine under: wasp_101.zip (0.587Mb)

Included is:
1. Wasp_101-bmi2-x64.exe (with bit manipulation instructions = BMI2)
2. Wasp_101-x64.exe (with no bit manipulation instructions for older CPU's) ... NEW
3. Wasp_101-w32.exe (compiled for athlon-64, should also work with Pentium 3 or greater Intel processors)
4. release-notes.txt
5. The two main logos by Wilhelm Hudetz

Thanks John

 

June 23rd, 2016:

FCP Rating List ended with 2740 Elo
Wasp 1.01 x64

  No bigger statistics differents after 100% compared to the statistics I added before after 50%.
  The move-statistics are very exactly!

   #  Player         :      Elo  Games  Score%   won  draw  lost  Points  Draw%  Error
  35. Wasp 1.01 x64  :  2738.10   1225    38.7   197   555   473   474.5   45.3  14.87
  35. Wasp 1.01 x64  :  2740.53   2450    39.1   409  1100   941   959.0   44.9  10.10

 

  Move number 020-059
  -------------------
  161 of 1.225 games
  29. Wasp 1.01 x64  :  2793.24    161    47.2    18   116    27    76.0   72.0  29.09  + 55.14 Elo!
  323 of 2.450 games
  25. Wasp 1.01 x64  :  2810.14    323    49.4    45   229    49   159.5   70.9  21.20  + 69.61 Elo!

  
  Move number 060-079
  -------------------
  387 of 1.225 games
  35. Wasp 1.01 x64  :  2709.06    387    35.7    83   110   194   138.0   28.4  32.37  - 29.04 Elo
  792 of 2.450 games
  36. Wasp 1.01 x64  :  2702.00    792    35.2   166   225   401   278.5   28.4  22.10  - 38.53 Elo


  Move number 080-099
  -------------------
  304 of 1.225 games
  40. Wasp 1.01 x64  :  2680.97    304    31.1    60    69   175    94.5   22.7  36.03  - 57.13 Elo
  633 of 2.450 games
  40. Wasp 1.01 x64  :  2690.85    633    32.9   131   155   347   208.5   24.5  24.90  - 49.68 Elo
  
  Move number 100-299
  -------------------
  373 of 1.225 games
  36. Wasp 1.01 x64  :  2784.97    373    44.5    36   260    77   166.0   69.7  20.39  + 46,87 Elo!
  702 of 2.450 games
  35. Wasp 1.01 x64  :  2785.58    702    44.5    67   491   144   312.5   69.9  14.03  + 45,05 Elo!

 

June 22nd, 2016:

John sent a 32-bit version. For the moment no interest for MAC / Android.

John wrote:
Attached is the 32-bit version of Wasp 1.01. It is compiled for athlon-64, but I think it should also work with Pentium 3 or greater Intel processors.  It works on my old Intel Core2 system with a 32 bit version of Windows.  But I don't have any actual 32 bit computers to try it on.  On my 64 bit i5-4690K it runs about 35% slower than the 64-bit version of 1.01 with bit manipulation instructions.

 

June 21st, 2016:

Wasp 1.01 x64 released!

John Stanback sent the release-notes.txt ... wow 37 years chess programming, read it (interesting for chess programmers & non-programmers)!

I have to wait of the end of the test-run!
In my main interest is to create *.epd test-suites. I will used the own material for searching good and bad positions from the produced games. The database I created for FCP Rating List is interesting for each participant chess programmer. It will be a nice example what we can do with the produced games. I am sure that databases in *.epd helps chess programmers a lot.

The test-run vs. a lot of different opponents is very nice:
It seems Wasp don't like: Hakkapeliitta, Texel, Arasan, Gaviota & Disaster Area. In my opinion the most interesting sparring partners for Wasp.

 

June 21st, 2016:

Copyright information to Zarkov / Wasp added ...
I updated my Impressum site.

 

June 21st, 2016:

After 50% of my test-run ...

It's absolutely clear: Wasp 1.01 x64 is very strong in endgames without using endgame databases. The playing style is great because all is totally "human like". Not easy to beat the engine after opening book moves. Sometimes very danger after opening book moves. Weak point is the late middlegame / transition into endgame. Same weak points humans have if we compared with the strongest available chess programs today. Often in endgames Wasp have problems with draw detection (can be see in eval). So I have at first good material / ideas for my beta test. Elo is very stable with ~ 2730 - 2740. now After 1.225 games no "lost on time" or other technically problems. In my humble opinion the perfect program for self-playing and much more important ... no other engine have such a style!!

Have a look here ...
If you working on such statistics with the strongest chess players you can see it in self work too ... the "Human like style".

  After 1.225 of 2.450 games = 50%
  --------------------------------
   #  Player         :      Elo  Games  Score%   won  draw  lost  Points  Draw%  Error
  35. Wasp 1.01 x64  :  2738.10   1225    38.7   197   555   473   474.5   45.3  14.87

   Move number 020-059
  -------------------
  29. Wasp 1.01 x64  :  2793.24    161    47.2    18   116    27    76.0   72.0  29.09  + 55.14 Elo!
  Stronger after opening book moves!

  Move number 060-079
  -------------------
  35. Wasp 1.01 x64  :  2709.06    387    35.7    83   110   194   138.0   28.4  32.37  - 29.04 Elo
  Lost strength in late middlegame.

  Move number 080-099
  -------------------
  40. Wasp 1.01 x64  :  2680.97    304    31.1    60    69   175    94.5   22.7  36.03  - 57.13 Elo
  Lost strength transition into endgame.

  Move number 100-299
  -------------------
  36. Wasp 1.01 x64  :  2784.97    373    44.5    36   260    77   166.0   69.7  20.39  + 46,87 Elo!
  And won the strength back in endgame! 

That's Zarkov and the main reason I like the work by John all the time. John never changed the own face Zarkov had and now Wasp is really 300 Elo stronger as Zarkov! I asked John to added again in next versions more UCI options for building other playing styles, like the older *.pst file from the Zarkov 4.1 Windows version. John have interest to do it later!

 

June 20th, 2016:

John is working on the readme file for the first final release.

John wrote:
I am working on the readme file that describes a few of the techniques used in Wasp.  There is nothing new or unusual compared to other programs, but it has it's own combination of search techniques and evaluation parameters which make it unique.

 

June 18th, 2016:

Wilhelm Hudetz sent indeed wonderful logos!
Logos with the USA & Kentucky flag and of course logos with a "Wasp"!

... our favorite!

     

 

June 18th, 2016:

Wasp 1.01 x64 test-run started!

 

June 18th, 2016:

Older Zarkov versions ...

zarkov-older-versions.zip (5.202Mb): MS DOS 2.x / 3.0 versions, MS Windows 4.1 version, 30 Winboard versions!!

Results from my older SWCR Rating List (40 moves in 10 minutes repeatedly, ponder = on, Intel® Core™ Q9550  hardware)

Shredder 12 x64  : 2800 after 9940 games
Zarkov 6.44 w32  : 2435 after 2340 games

The successor Wasp 1.01 x64 is ~ 300 Elo stronger!

 

June 17th, 2016:

I contact Wilhelm Hudetz. Wilhelm is working with me in the older Arena team and created nice engine logos. I hope and wish me that Wilhelm have time to create some nice logos.

 

June 16th, 2016:

My first contact after a long time with John Stanback!
Reason was a discussion in TalkChess to the topic "xb / uci chronology" I powered WB / UCI in the past a long time.

Have a look under: Computer-Chess-Wiki (Ron Murawski & Norbert Raimund Leisner do here really a fantastic work for all of us)
Have a look under: RWBC-chess by Günther Simon (very strong computer chess history site).
Have a look under: Chess-Programming-Wiki (Expert Gerd Isenberg is working all the time ...)

John wrote:
I've been planning to make Wasp public, but I keep hoping to make a big strength jump before I do.  But improvements are very hard to come by, so I guess version 1.01 is good enough for now.  Wasp only uses a single core, maybe I'll try to add support for multiple cores later.  Also, it doesn't use endgame tables.  This is compiled for fairly recent Intel 64 bit CPU's.

After the results John speaking about I think Elo could be around ~ 2750 for my FCP Rating List. John sent me Wasp 1.01 x64 for a first test. The test-run will be start June 18th, 2016.

 

Flag Counter