SPCC: Testruns of Rebel EAS 2 and Rebel Extreme finished

Posts : 160 Join date : 2022-03-01 Location : Berlin

The testruns of Rebel EAS 2.0 and Rebel Extreme are finished. Only for my full Ratinglist, because both are too weak to enter the UHO-Top15 ratinglist.

https://www.sp-cc.de/files/uho_full_list.txt

Code:: 50 Viridithas 13.0 avx2 : 3589 3 3 22000 38.1% 3678 49.0%
51 CSTal 2.0 avx2 : 3588 2 2 64000 47.4% 3609 50.4%
52 Rebel 16.3 avx2 : 3585 3 3 31000 43.0% 3637 49.8%
53 Uralochka 3.41a avx2 : 3585 3 3 26000 41.1% 3650 48.8%
54 Obsidian 9.0 avx2 : 3582 4 4 17000 43.9% 3628 49.3%
55 Titan 1.0 avx2 : 3579 4 4 20000 45.2% 3607 45.9%
56 Clover 6.1 avx2 : 3578 3 3 38000 44.6% 3619 49.7%
57 Rebel EAS 2.0 avx2 : 3575 4 4 14000 56.5% 3529 45.8% (NEW)
58 Koivisto 9.2 avx2 : 3574 2 2 58000 47.6% 3590 47.2%
59 Clarity 7.2.0 avx2 : 3572 4 4 18000 44.4% 3606 45.6%
60 Alexandria 6.0 avx2 : 3565 4 4 22000 41.5% 3630 48.0%
61 Caissa 1.13.1 avx2 : 3561 4 4 21000 42.8% 3616 49.4%
62 Velvet 8.0.0 normal : 3559 4 4 15000 33.6% 3682 46.4%
63 Clover 6 avx2 : 3558 3 3 34000 47.9% 3572 47.3%
64 Rebel EAS avx2 : 3554 3 3 44000 44.0% 3600 49.0%
65 Lizard 10.4 avx2 : 3553 4 4 14000 34.1% 3674 49.2%
66 Seer 2.7.0 avx2 : 3548 3 3 22000 41.8% 3610 48.5%
67 RofChade 3.1 avx2 : 3547 2 2 51000 44.2% 3590 46.0%
68 Viridithas 12.0 avx2 : 3547 4 4 16000 37.2% 3644 47.6%
69 Caissa 1.12 avx2 : 3544 4 4 22000 48.7% 3550 47.0%
70 Starzix 5.0 avx2 : 3544 4 4 18000 42.2% 3598 43.9%
71 Velvet 8.0.0 risky : 3541 4 4 15000 31.4% 3684 42.1%
72 Revenge 3.0 avx2 : 3538 3 3 30000 41.0% 3607 48.8%
73 Uralochka 3.40a avx2 : 3537 3 3 25000 39.3% 3617 47.3%
74 Igel 3.5.0 popavx2 : 3533 3 3 29000 44.8% 3570 45.8%
75 Obsidian 8.0 avx2 : 3532 4 4 18000 45.1% 3565 44.9%
76 Velvet 7.3.0 avx2 : 3526 4 4 16000 34.0% 3648 44.1%
77 Alexandria 5.1.0 avx2 : 3515 5 5 14000 35.7% 3624 46.1%
78 Seer 2.6.0 avx2 : 3507 3 3 28000 41.4% 3572 44.0%
79 Arasan 24.2.1 avx2 : 3496 5 5 15000 29.2% 3658 43.5%
80 Alexandria 5.0.0 avx2 : 3491 4 4 18000 43.0% 3541 43.4%
81 Viridithas 11.0.0 avx2 : 3488 4 4 20000 42.2% 3546 44.0%
82 Velvet 7.0.0 avx2 : 3484 4 4 16000 30.7% 3633 43.1%
83 Stockfish final HCE : 3479 4 4 18000 41.2% 3543 37.9%
84 Slow Chess 2.9 avx2 : 3455 4 4 18000 39.1% 3537 42.5%
85 Rebel Extreme avx2 : 3411 4 4 14000 34.0% 3529 39.3% (NEW)
86 Revenge 1.0 avx2 : 3365 5 5 15000 18.3% 3637 30.3%
87 Patricia 3.1 avx2 : 3218 6 6 14000 14.7% 3529 21.1%
88 Patricia 3.0 avx2 : 3215 6 6 14000 14.5% 3529 21.2%

Here the EAS-Ratinglist (Top10):

Code::    bad avg.win
Rank EAS-Score sacs shorts draws moves Engine/player
-------------------------------------------------------------------
   1 429481 51.61% 38.03% 05.49% 66 Patricia 3.1 avx2
   2 384942 52.55% 35.77% 05.19% 69 Patricia 3.0 avx2
   3 250682 37.88% 23.88% 08.79% 75 Rebel Extreme avx2 (NEW)
   4 201351 34.75% 22.22% 10.27% 73 Velvet 8.0.0 risky
   5 197919 31.18% 29.46% 17.09% 71 Revenge 1.0 avx2
   6 192881 21.25% 26.18% 09.40% 70 Stockfish 16 230630
   7 190148 21.94% 24.53% 08.88% 70 Stockfish 16.1 240224
   8 186267 25.24% 24.69% 13.42% 72 Velvet 7.0.0 avx2
   9 184447 24.86% 20.78% 10.07% 75 Rebel EAS 2.0 avx2 (NEW)
  10 181381 20.24% 25.04% 10.47% 70 Stockfish 15.1 221204

I find these results of the 2 new Rebels quite disappointing, especially when comparing them with Velvet 8 (normal/risky):
Rebel Extreme gained 66000 EAS-points compared to Rebel EAS 2.0, but lost -164 Celo.
Velvet 8 risky (EAS:201351) gained 104000 EAS-points compared to Velvet 8 normal(EAS:96753) (=doubled the EAS-score!), but lost only -18 Celo !!!

Code::    bad avg.win
Rank EAS-Score sacs shorts draws moves Engine/player
-------------------------------------------------------------------
   4 201351 34.75% 22.22% 10.27% 73 Velvet 8.0.0 risky (3541 Celo)
  37 96753 11.66% 18.36% 17.32% 76 Velvet 8.0.0 normal (3559 Celo)

Thank for testing the 2 new Rebel's in the first place.

Contrary to you I am not disappointed at all.

Rebel-EAS-2.0 in comparison with 16.3 doubled its EAS score with hardly any elo loss, it's a signal I am on the right track.

Rebel-Extreme, second after Patricia and about 200 elo stronger, perfect start.

I have enough ideas left for both concepts.

BTW, in your test you let Rebel play against 2 x Patricia, that means all 3 lose EAS points, see my post elsewhere. If you remove the games against the 2 x Patricia's I am pretty sure the EAS will be higher.

BTW-2, I can't find the games of the 2 Rebel's on your website, do I need new glasses?

Posts : 160 Join date : 2022-03-01 Location : Berlin

Admin wrote:: BTW, in your test you let Rebel play against 2 x Patricia, that means all 3 lose EAS points, see my post elsewhere. If you remove the games against the 2 x Patricia's I am pretty sure the EAS will be higher.

??? Completely wrong. Of course, these aggressive engines are never tested against each other in my testings (sorry, that I did not mention this earlier):

Here the single stats of Rebel:

Code:: 57 Rebel EAS 2.0 avx2 : 3575 14000 (+4702,=6412,-2886), 56.5 %

Clarity 7.2.0 avx2 : 1000 (+258,=476,-266), 49.6 %
Starzix 5.0 avx2 : 1000 (+305,=474,-221), 54.2 %
Titan 1.0 avx2 : 1000 (+263,=467,-270), 49.6 %
Obsidian 8.0 avx2 : 1000 (+313,=485,-202), 55.5 %
Stockfish final HCE : 1000 (+429,=405,-166), 63.1 %
Alexandria 5.0.0 avx2 : 1000 (+408,=426,-166), 62.1 %
RofChade 3.1 avx2 : 1000 (+320,=475,-205), 55.8 %
Viridithas 11.0.0 avx2 : 1000 (+408,=457,-135), 63.6 %
Slow Chess 2.9 avx2 : 1000 (+427,=457,-116), 65.5 %
Seer 2.6.0 avx2 : 1000 (+369,=427,-204), 58.3 %
Igel 3.5.0 popavx2 : 1000 (+345,=450,-205), 57.0 %
Caissa 1.12 avx2 : 1000 (+319,=469,-212), 55.4 %
Clover 6 avx2 : 1000 (+271,=492,-237), 51.7 %
Koivisto 9.2 avx2 : 1000 (+267,=452,-281), 49.3 %

Code:: 85 Rebel Extreme avx2 : 3411 14000 (+2014,=5496,-6490), 34.0 %

Clarity 7.2.0 avx2 : 1000 (+ 96,=378,-526), 28.5 %
Starzix 5.0 avx2 : 1000 (+128,=384,-488), 32.0 %
Titan 1.0 avx2 : 1000 (+112,=341,-547), 28.3 %
Obsidian 8.0 avx2 : 1000 (+115,=429,-456), 33.0 %
Stockfish final HCE : 1000 (+206,=401,-393), 40.6 %
Alexandria 5.0.0 avx2 : 1000 (+184,=391,-425), 38.0 %
RofChade 3.1 avx2 : 1000 (+124,=410,-466), 32.9 %
Viridithas 11.0.0 avx2 : 1000 (+192,=413,-395), 39.9 %
Slow Chess 2.9 avx2 : 1000 (+231,=416,-353), 43.9 %
Seer 2.6.0 avx2 : 1000 (+180,=375,-445), 36.8 %
Igel 3.5.0 popavx2 : 1000 (+135,=428,-437), 34.9 %
Caissa 1.12 avx2 : 1000 (+121,=394,-485), 31.8 %
Clover 6 avx2 : 1000 (+ 90,=380,-530), 28.0 %
Koivisto 9.2 avx2 : 1000 (+100,=356,-544), 27.8 %

Posts : 160 Join date : 2022-03-01 Location : Berlin

Admin wrote:: BTW-2, I can't find the games of the 2 Rebel's on your website, do I need new glasses?

Only in my full Ratinglist. Too weak for my UHO-Top15 Ratinglist:

https://www.sp-cc.de/files/uho_full_list.txt

Posts : 160 Join date : 2022-03-01 Location : Berlin

Admin wrote:: Thank for testing the 2 new Rebel's in the first place.

Contrary to you I am not disappointed at all.

Rebel-EAS-2.0 in comparison with 16.3 doubled its EAS score with hardly any elo loss, it's a signal I am on the right track.

Rebel-Extreme, second after Patricia and about 200 elo stronger, perfect start.

I have enough ideas left for both concepts.

OK, Rebel EAS 2.0 compared with Rebel 16.3 looks better, indeed. But, what M.Honert (Velvet) achieved is very, very impressing. He lost less than -20 Celo and gained 100k EAS-points.

The Patricia author just claimed on discord, the new Patricia dev is around +200 Elo stronger than V3.1 (would mean the same strength as Rebel Extreme), without loosing EAS (still around 400k). Of course, this is only a claim. But in the past, his predictions fitted quite well to my following testings.

Posts : 160 Join date : 2022-03-01 Location : Berlin

On discord (Stockfish engines dev) the Viridithas author made this, could be interesting for you, I presume. Each yellow X is one engine of my full EAS-ratinglist. Great to see, that EAS becomes interesting for more and more people in our computerchess-bubble. Especially interesting, because Viridithas has a very low EAS-scoring...

https://en.wikipedia.org/wiki/Pareto_front

Current Pareto frontier for engine aggressiveness / strength (good news: Rebel Extreme is one of the engines on the border-line (but only until the new Patricia is released?!?), but Rebel EAS 2 is not. So, the work must go on...)

SPCC: Testruns of Rebel EAS 2 and Rebel Extreme finished Image

SPCC: Testruns of Rebel EAS 2 and Rebel Extreme finished Image

pohl4711 wrote:

Admin wrote:: BTW, in your test you let Rebel play against 2 x Patricia, that means all 3 lose EAS points, see my post elsewhere. If you remove the games against the 2 x Patricia's I am pretty sure the EAS will be higher.

??? Completely wrong. Of course, these aggressive engines are never tested against each other in my testings (sorry, that I did not mention this earlier):

Yes you did say that, ok, no problem.

pohl4711 wrote:

Admin wrote:: BTW-2, I can't find the games of the 2 Rebel's on your website, do I need new glasses?

Only in my full Ratinglist. Too weak for my UHO-Top15 Ratinglist:

https://www.sp-cc.de/files/uho_full_list.txt

Can you please upload the games of the 2 Rebel's for me, it's important for my evaluation.

pohl4711 wrote:

Admin wrote:: Thank for testing the 2 new Rebel's in the first place.

Contrary to you I am not disappointed at all.

Rebel-EAS-2.0 in comparison with 16.3 doubled its EAS score with hardly any elo loss, it's a signal I am on the right track.

Rebel-Extreme, second after Patricia and about 200 elo stronger, perfect start.

I have enough ideas left for both concepts.

OK, Rebel EAS 2.0 compared with Rebel 16.3 looks better, indeed. But, what M.Honert (Velvet) achieved is very, very impressing. He lost less than -20 Celo and gained 100k EAS-points.

The Patricia author just claimed on discord, the new Patricia dev is around +200 Elo stronger than V3.1 (would mean the same strength as Rebel Extreme), without loosing EAS (still around 400k). Of course, this is only a claim. But in the past, his predictions fitted quite well to my following testings.

Wow........ Very Happy

Posts : 160 Join date : 2022-03-01 Location : Berlin

Admin wrote:: Can you please upload the games of the 2 Rebel's for me, it's important for my evaluation.

https://drive.google.com/file/d/1-NuJ9YyGZb24Z0KoKQ8Gzw6Fnx9a-hqD/view?usp=sharing

(I will delete this file in some days, because I need the space on my GoogleDrive)

The games of Rebel EAS 2.0 and Rebel Extreme can be downloaded anytime later, when downloading my UHO_Ratinglist_Archive005 zip-file:
https://drive.google.com/drive/folders/166wG-5OZKc4RL5qlbKHYd3yjfrdepJOv?usp=drive_link
Here the games still have all comments/evals...

(In the UHO_Full_Ratinglist zip-file, the games are included, too, but without any comments/evals)

Posts : 160 Join date : 2022-03-01 Location : Berlin

Admin wrote:

pohl4711 wrote:

Admin wrote:: BTW, in your test you let Rebel play against 2 x Patricia, that means all 3 lose EAS points, see my post elsewhere. If you remove the games against the 2 x Patricia's I am pretty sure the EAS will be higher.

??? Completely wrong. Of course, these aggressive engines are never tested against each other in my testings (sorry, that I did not mention this earlier):

Yes you did say that, ok, no problem.

And (additionally), both new Rebels played the exact same opponents as Patricia 3.0/3.1. So the results are perfectly comparable. New Patricia will play the same opponents, too, whenever she will be released.

pohl4711 wrote:

Admin wrote:

pohl4711 wrote:

Admin wrote:: BTW, in your test you let Rebel play against 2 x Patricia, that means all 3 lose EAS points, see my post elsewhere. If you remove the games against the 2 x Patricia's I am pretty sure the EAS will be higher.

??? Completely wrong. Of course, these aggressive engines are never tested against each other in my testings (sorry, that I did not mention this earlier):

Yes you did say that, ok, no problem.

And (additionally), both new Rebels played the exact same opponents as Patricia 3.0/3.1. So the results are perfectly comparable. New Patricia will play the same opponents, too, whenever she will be released.

Downloaded the games, thank you.

Code:: No. Name Win Draw Loss Unf. Score Games %
-------------------------------------------------------------------
  1 Rebel EAS 2.0 avx2    +4702 =6412 -2886 *0 7908.0 14000 56.5%
  2 Koivisto 9.2 avx2 +281 =452 -267 *0 507.0 1000 50.7%
  3 Clarity 7.2.0 avx2 +266 =476 -258 *0 504.0 1000 50.4%
  4 Titan 1.0 avx2 +270 =467 -263 *0 503.5 1000 50.4%
  5 Clover 6 avx2 +237 =492 -271 *0 483.0 1000 48.3%
  6 Starzix 5.0 avx2 +221 =474 -305 *0 458.0 1000 45.8%
  7 Caissa 1.12 avx2 +212 =469 -319 *0 446.5 1000 44.6%
  8 Obsidian 8.0 avx2 +202 =485 -313 *0 444.5 1000 44.5%
  9 RofChade 3.1 avx2 +205 =475 -320 *0 442.5 1000 44.2%
10 Igel 3.5.0 popavx2 +205 =450 -345 *0 430.0 1000 43.0%
11 Seer 2.6.0 avx2 +204 =427 -369 *0 417.5 1000 41.8%
12 Alexandria 5.0.0 avx2 +166 =426 -408 *0 379.0 1000 37.9%
13 Stockfish final HCE +166 =405 -429 *0 368.5 1000 36.9%
14 Viridithas 11.0.0 avx2 +135 =457 -408 *0 363.5 1000 36.4%
15 Slow Chess 2.9 avx2 +116 =457 -427 *0 344.5 1000 34.5%

What springs to mind is that you have chosen an almost exact elo pool as I use, very nice. Of course your time control (as opposed to mine : bullet) as expected lowers the EAS, but fortunately not too much. The same applies for playing games till mate, while I don't, has an effect on the game length and thus also on the EAS.

Good testing, thank you.

Posts : 160 Join date : 2022-03-01 Location : Berlin

Admin wrote:: What springs to mind is that you have chosen an almost exact elo pool as I use, very nice. Of course your time control (as opposed to mine : bullet) as expected lowers the EAS, but fortunately not too much. The same applies for playing games till mate, while I don't, has an effect on the game length and thus also on the EAS.

Good testing, thank you.

This could be interesting for you and for your testing and for comparison:

https://drive.google.com/file/d/1QcpBTjj2HVkiqt4wRXpQN4E0wOgkF1wc/view?usp=sharing

I wrote a small readme:
Alexander 2.0 is a fork of ShashChess, which is a fork of Stockfish.
https://github.com/amchess/Alexander

But Alexander uses only the old HCE of Stockfish (up to V11), and allows to change
some of the HCE evaluation-parameters in an avatar-file...

Use UCI option "Avatar File" to chose the avatar-file
Additionally set UCI option "High Tal" to true (default false)

alew 100 percent setting was made by Peter Martan, the others by Stefan Pohl (SPCC)

The alew setting sets some of the eval-parameters (which all have 100 as default) down:

alew 100 percent: Some parameters are dropped from 100 to 50 and some from 100 to 70.

alew 90 percent: Some parameters are dropped from 100 to 55 and some from 100 to 73 (-10% less drop,
compared to alew 100 percent)

alew 80 percent: Some parameters are dropped from 100 to 60 and some from 100 to 76 (-20% less drop,
compared to alew 100 percent)

alew 70 percent: Some parameters are dropped from 100 to 65 and some from 100 to 79 (-30% less drop,
compared to alew 100 percent)

alew 100 percent is the weakest, but most aggressive setting. The lower the percent-number of the
other alew-settings, the higher the strength, but the lower the aggressiveness.

Alexander 2.0 alew 100 percent is around -130 Celo weaker than Stockfish final HCE and around the
strength of Revenge 1.0.

So, here you have a superaggressive not-neural-HCE-(Stockfish-)Engine, which can be adjusted in aggressiveness and strength, just by choosing one of the 4 avatar-files !!!

I started a testrun of Alexander 2 alew 100 (same 14 opponents as for the Rebel-testruns and Patricia-testruns), and after 4000 games, the EAS-Score is nearly 400.000 (!!!)

So, right now, the Pareto chart would look like this (if the results are stable until the testrun is finished):

SPCC: Testruns of Rebel EAS 2 and Rebel Extreme finished EAS_Pareto_frontier_Alex_alew

SPCC: Testruns of Rebel EAS 2 and Rebel Extreme finished EAS_Pareto_frontier_Alex_alew

Posts : 160 Join date : 2022-03-01 Location : Berlin

There is a new release of Alexander (Santiago), I made a new download, with old and new binaries:
https://drive.google.com/file/d/1x34IY3ldnmpEeuOAdxoXk76cQOFFQsCp/view?usp=sharing

And I restarted the testrun with alew100 setting and the new binary.

Posts : 160 Join date : 2022-03-01 Location : Berlin

pohl4711 wrote:: There is a new release of Alexander (Santiago), I made a new download, with old and new binaries:
https://drive.google.com/file/d/1x34IY3ldnmpEeuOAdxoXk76cQOFFQsCp/view?usp=sharing

And I restarted the testrun with alew100 setting and the new binary.

BUG ALERT ! The new version (Santiago) plays just crap. Lost 29 out of 30 games, 1 draw...

So, all back to start (Alexander 2.0):
https://drive.google.com/file/d/1QcpBTjj2HVkiqt4wRXpQN4E0wOgkF1wc/view?usp=sharing

Posts : 5 Join date : 2022-12-01

pohl4711 wrote:: The testruns of Rebel EAS 2.0 and Rebel Extreme are finished. Only for my full Ratinglist, because both are too weak to enter the UHO-Top15 ratinglist.

https://www.sp-cc.de/files/uho_full_list.txt

Rebel 16.3 still ahead. No improvements after 1 year

[code]
50 Viridithas 13.0 avx2 : 3589 3 3 22000 38.1% 3678 49.0%
51 CSTal 2.0 avx2 : 3588 2 2 64000 47.4% 3609 50.4%
52 Rebel 16.3 avx2 : 3585 3 3 31000 43.0% 3637 49.8%
53 Uralochka 3.41a avx2 : 3585 3 3 26000 41.1% 3650 48.8%
54 Obsidian 9.0 avx2 : 3582 4 4 17000 43.9% 3628 49.3%
55 Titan 1.0 avx2 : 3579 4 4 20000 45.2% 3607 45.9%
56 Clover 6.1 avx2 : 3578 3 3 38000 44.6% 3619 49.7%
57 Rebel EAS 2.0 avx2 : 3575 4 4 14000 56.5% 3529 45.8% (NEW)
58 Koivisto 9.2 avx2 : 3574 2 2 58000 47.6% 3590 47.2%
59 Clarity 7.2.0 avx2 : 3572 4 4 18000 44.4% 3606 45.6%
60 Alexandria 6.0 avx2 : 3565 4 4 22000 41.5% 3630 48.0%
61 Caissa 1.13.1 avx2 : 3561 4 4 21000 42.8% 3616 49.4%
62 Velvet 8.0.0 normal : 3559 4 4 15000 33.6% 3682 46.4%
63 Clover 6 avx2 : 3558 3 3 34000 47.9% 3572 47.3%
64 Rebel EAS avx2 : 3554 3 3 44000 44.0% 3600 49.0%
65 Lizard 10.4 avx2 : 3553 4 4 14000 34.1% 3674 49.2%

Posts : 24 Join date : 2022-10-23

pohl4711 wrote:: So, all back to start (Alexander 2.0):

Thanks for testing "my" .avt- setting, told Andrea Manzo about a bug in Alexander Santiago you noticed, he answered he fixed one (a bug) already, but better wait till I've tried a new one download, he won't have much time at the moment of his own, regards
Peter.

Edit: Signature in forum here seems not to work for me

Edit, edit: A first match between Alexander Santiago (re-downloaded just after starting this posting) and 2.0 in CuteChess GUI with default settings of both, 3'+1" single thread, UHO 2024 900-990cp 6mvs, ran smooth without any disconnets nor losses on time:

Score of AlexanderSantiago vs Alexander2.0: 280 - 63 - 157 [0.717]
Elo difference: 161.5 +/- 26.5, LOS: 100.0 %, DrawRatio: 31.4 %
500 of 500 games finished.

Can send you the games by mail, if you want me to, but I guess, they aren't of any special interest, are they?

Posts : 160 Join date : 2022-03-01 Location : Berlin

peter wrote:

pohl4711 wrote:: So, all back to start (Alexander 2.0):

Thanks for testing "my" .avt- setting, told Andrea Manzo about a bug in Alexander Santiago you noticed, he answered he fixed one (a bug) already, but better wait till I've tried a new one download, he won't have much time at the moment of his own, regards
Peter.

Edit: Signature in forum here seems not to work for me

Edit, edit: A first match between Alexander Santiago (re-downloaded just after starting this posting) and 2.0 in CuteChess GUI with default settings of both, 3'+1" single thread, UHO 2024 900-990cp 6mvs, ran smooth without any disconnets nor losses on time:

Score of AlexanderSantiago vs Alexander2.0: 280 - 63 - 157 [0.717]
Elo difference: 161.5 +/- 26.5, LOS: 100.0 %, DrawRatio: 31.4 %
500 of 500 games finished.

Can send you the games by mail, if you want me to, but I guess, they aren't of any special interest, are they?

I redownloaded the binaries, they are the same like this morning, no change. So still buggy.

The games looked quite normal, but all games were lost (except 1 draw)

[Event "UHO Ratinglist"]
[Site "?"]
[Date "2024.10.18"]
[Round "1"]
[White "AlexSantiago alew100"]
[Black "Caissa 1.12 avx2"]
[Result "0-1"]
[ECO "A21"]
[GameDuration "00:06:31"]
[GameEndTime "2024-10-18T08:15:13.550 Mitteleuropäische Sommerzeit"]
[GameStartTime "2024-10-18T08:08:41.781 Mitteleuropäische Sommerzeit"]
[Opening "English, Kramnik-Shirov counterattack"]
[PlyCount "110"]
[TimeControl "180+1"]

1. c4 {book} e5 {book} 2. Nc3 {book} Bb4 {book} 3. Nd5 {book} Be7 {book}
4. d4 {book} d6 {book} 5. e4 {book} c6 {book} 6. Nxe7 {book} Qxe7 {book}
7. Ne2 {+0.11/24 9.1s} Nf6 {-1.21/25 9.6s} 8. f3 {+0.10/25 4.6s}
d5 {-0.57/28 8.3s} 9. cxd5 {+0.11/27 6.4s} cxd5 {-0.46/25 1.2s}
10. dxe5 {+0.08/28 5.4s} Qxe5 {-0.70/26 1.7s} 11. exd5 {+0.12/28 6.8s}
O-O {-0.55/29 6.5s} 12. Qd4 {+0.10/25 4.4s} Qxd5 {-0.49/29 7.9s}
13. Qxd5 {+0.12/24 3.3s} Nxd5 {-0.51/26 1.0s} 14. Nd4 {+0.10/26 5.4s}
Nc6 {-0.53/26 6.5s} 15. Nxc6 {+0.12/24 3.0s} bxc6 {-0.36/26 4.9s}
16. a3 {+0.11/30 26s} a5 {-0.15/27 6.7s} 17. Be2 {+0.12/25 4.0s}
Ba6 {0.00/27 4.7s} 18. Bd1 {+0.11/24 3.3s} Bd3 {0.00/28 6.0s}
19. h4 {+0.07/27 16s} c5 {+0.35/27 7.6s} 20. Kf2 {+0.06/26 8.5s}
Rfe8 {+0.14/26 6.5s} 21. Ra2 {+0.03/25 6.7s} c4 {+0.84/26 11s}
22. Rg1 {0.00/30 5.0s} Rab8 {+0.96/25 6.9s} 23. g4 {0.00/33 13s}
f6 {+1.28/23 5.4s} 24. g5 {0.00/28 3.7s} f5 {+1.40/23 5.1s}
25. Kg3 {0.00/33 2.7s} g6 {+1.91/24 4.9s} 26. Rh1 {0.00/32 5.9s}
Kf7 {+2.25/22 5.9s} 27. Rg1 {-0.05/27 8.8s} Re7 {+2.25/23 4.9s}
28. Rh1 {-0.05/27 2.0s} Re6 {+2.63/23 3.6s} 29. Ra1 {-0.05/28 14s}
Ke7 {+2.51/24 4.8s} 30. Ra2 {-0.09/28 13s} Kd6 {+2.84/22 3.0s}
31. a4 {-0.12/27 10s} Rbe8 {+5.58/24 4.2s} 32. b3 {-0.69/27 12s}
c3 {+6.06/20 1.6s} 33. Bc2 {-0.48/18 0.70s} Re2 {+6.89/18 0.65s}
34. Bxd3 {-0.47/19 0.58s} Rxa2 {+7.44/18 0.92s} 35. h5 {-0.66/22 3.5s}
Kc5 {+7.83/18 0.67s} 36. Bb5 {-0.67/19 0.68s} Re5 {+7.98/16 0.74s}
37. h6 {-0.73/20 0.97s} c2 {+8.84/17 0.96s} 38. Bd2 {-0.95/17 1.7s}
f4+ {+11.01/21 1.7s} 39. Kg4 {-0.93/16 0.15s} Kd4 {+11.37/21 1.3s}
40. Rc1 {-1.20/22 1.8s} Rf5 {+12.45/23 2.3s} 41. Kh4 {-1.25/16 0.54s}
Ne3 {+13.82/22 1.5s} 42. b4 {-1.69/18 1.5s} axb4 {+19.91/21 2.4s}
43. Bc6 {-2.41/18 1.1s} b3 {+22.83/19 1.2s} 44. Be4 {-3.03/17 0.62s}
Rxa4 {+25.06/17 1.8s} 45. Be1 {-3.34/20 1.3s} Rc5 {+29.71/20 1.3s}
46. Bf2 {-3.64/18 0.50s} Rac4 {+32.31/18 1.5s} 47. Be1 {-4.90/19 1.5s}
b2 {+44.50/21 0.94s} 48. Bd2 {-5.56/16 0.50s} bxc1=Q {+48.65/21 1.6s}
49. Bxc1 {-4.39/16 0.44s} Rb5 {+49.91/21 1.6s} 50. Bxg6 {-M14/38 1.0s}
Rb1 {+53.65/21 0.90s} 51. Bf5 {-M16/23 1.9s} Rxc1 {+M9/37 1.0s}
52. Bh3 {-M8/84 0.32s} Rh1 {+M7/65 0.61s} 53. g6 {-M6/245 0.11s}
Rc5 {+M5/255 0.25s} 54. g7 {-M4/245 0.011s} Rxh3+ {+M3/255 0.011s}
55. Kxh3 {-M2/245 0.008s} Rh5# {Black mates} 0-1

Posts : 24 Join date : 2022-10-23

Sorry, Stefan, only "bug" I found so far, is the name of some of the binaries in download reads Aleander instead of Alexander, e.g. this is so as for the bmi2- compile I used, but I simply renamed that (which I wouldn't have had to neither probably, because started in console the version- name was ok anyhow).
Didn't have more than the 500 games quoted above, but as told, no bug nor unexpected weakness to be seen there for me.
Caissa 1.12 maybe is too strong an opponent for a non- NNUE- engine like Alexander in handicap- mode anyhow still, but that I don't have to tell to you, I guess
Smile

One difference between your match and mine is the alew.avt you used and I didn't (this time), wanting to see the default- setting play against the same as for 2.0. Theoretically Santiago could have problems wit alew.avt, will give that a try now too.

Let me know, if you want to get my .pgn, if I yet find any bug in Alexander Santiago in follow- up tests, I'll report it too.

Posts : 24 Join date : 2022-10-23

To see, if the avatar- setting causes trouble, I had another one match Alexander Santiago- Alexander 2.0, this time both engines using the alew.avt- setting, again 3'+1" single thread, UHO 2024 900-990cp 6mvs:

Score of SantiagoAlew vs Alex2Alew: 161 - 79 - 260 [0.582]
Elo difference: 57.5 +/- 21.1, LOS: 100.0 %, DrawRatio: 52.0 %
500 of 500 games finished.

And with EAS- tool:

Code::    bad avg.win
Rank EAS-Score sacs shorts draws moves Engine/player
-------------------------------------------------------------------
   1 59380 06.83% 10.56% 20.77% 74 SantiagoAlew
   2 45200 05.06% 15.19% 24.23% 76 Alex2Alew
-------------------------------------------------------------------
*** Average length of all won games: 75 moves

What is Alexander Santiago?

Does it play in the upcoming ICGA tournament?

Posts : 222 Join date : 2021-08-28

yes and no. Shashchess is playing.

https://icga.org/?page_id=3957

https://github.com/amchess/Alexander

Posts : 24 Join date : 2022-10-23

No, that's ShashChess (playing in Santiago).
As Stefan wrote, Alexander is the "little brother" of ShashChess from same author (Andrea Manzo), not using NNUE but HCE and meant for human players to have (by UCI- Elo or) many eval- parameters adaptable as handicap- modes of special playing styles.

Edit: didn't see, you had edited your answer from yes to yes and no, while I was already typing too
Smile

» SPCC: Testrun of Rebel 16.1 finished
» SPCC: Testrun of Rebel 15.1a finished
» SPCC: Testrun of Torch 2 finished
» SPCC: Testrun of Torch 1 finished
» SPCC: Test of Pedone 3 Strength-parameter finished