SF PB (Stockfish)

Orgz · Post by **Orgz** » Mon Sep 11, 2023 10:10 pm

Orgz wrote: ↑Sat Jul 22, 2023 9:38 am Hope it doesn't offend you skynet that i posted here. i didnt want to open another sf pb post which is unnecessary. anyway

SF-POLY 190723
Same as the father Stockfish 190723 capable of using 4 bin books + a little search change. on my device the engine is better than the father. since engines behave differently on devices u'll see how he plays on yourS, thank you enjoy
https://pixeldrain.com/u/ax2F5s4a

TR

Knock knock Chess Friends

update SF-POLY 110923 2 bin books. books can be added in Raid style i.e. put bin book in the engine folder and rename to Poly1.bin / Poly2.bin.
https://pixeldrain.com/u/6svXdin3

Score of SF-POLY vs Stockfish 030923_sse41: 19 - 15 - 116 [0.513]
... SF-POLY playing White: 15 - 1 - 59 [0.593] 75
... SF-POLY playing Black: 4 - 14 - 57 [0.433] 75
... White vs Black: 29 - 5 - 116 [0.580] 150
Elo difference: 9.3 +/- 26.5, LOS: 75.4 %, DrawRatio: 77.3 %
150 of 150 games finished.

happy testing
TR

Skynet · Post by **Skynet** » Wed Sep 13, 2023 11:51 pm

SF-PB-110923 "Reformat some comments"
https://pixeldrain.com/u/VKRDjGMR

Skynet · Post by **Skynet** » Sun Sep 24, 2023 11:01 pm

SF-PB-250923 "Explicitly invoke shell".
https://pixeldrain.com/u/rFwt1Rdr
===============================
Note: Adrian, i wouldn't recommend you to use this version for book tests, because due to larger network on slower PC it gives a big regression, failed on single core ->
Failed 10+0.1 STC:
https://tests.stockfishchess.org/tests/ ... da89abab21
LLR: -2.92 (-2.94,2.94) <0.00,2.00>
Total: 13184 W: 3285 L: 3535 D: 6364 Elo -6.59
Ptnml(0-2): 85, 1662, 3334, 1440, 71

Failed 180+1.8 VLTC:
https://tests.stockfishchess.org/tests/ ... c881ea908e
LLR: -2.94 (-2.94,2.94) <0.00,2.00>
Total: 64248 W: 16224 L: 16374 D: 31650 Elo -0.81
Ptnml(0-2): 26, 6788, 18640, 6650, 20
=============================================================================
Passed with 8 threads (just a small gain) ->
Passed 60+0.6 th 8 VLTC SMP (STC bounds):
https://tests.stockfishchess.org/tests/ ... 74c2e541dc
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 90630 W: 23372 L: 23033 D: 44225 Elo +1.30
Ptnml(0-2): 13, 8490, 27968, 8833, 11

Passed 60+0.6 th 8 VLTC SMP:
https://tests.stockfishchess.org/tests/ ... da89abacdb
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 137804 W: 35764 L: 35276 D: 66764 Elo +1.23
Ptnml(0-2): 31, 13006, 42326, 13522, 17
With networks and patches like this, soon everyone will need a supercomputer to use Stockfish.

See ya in couple days on Outskirts, need to go now.

Post by **massimilianogoi** » Mon Sep 25, 2023 5:20 am

I don't know what your parameters of comparison are, but too few time has passed since your previous release, and if you refer about the Stockfish benchmarks... well, they must do it because it's their standard method. But a Stockfish derived engine has sense to be updated every three months or more, unless a bug is discovered in a network or an evident enhacement in performance has been done (i.e. 15-20 elo points more since the latest release).

This because the engine performance tends to flatten once associated with a book and on online competition with others (i.e. Playchess). There are people who even gets all draw games with the latest Komodo, a cheap processor and a good book.

Skynet wrote: ↑Sun Sep 24, 2023 11:01 pm ...
Note: Adrian, i wouldn't recommend you to use this version for book tests, because due to larger network on slower PC it gives a big regression, failed on single core ->
Failed 10+0.1 STC:
https://tests.stockfishchess.org/tests/ ... da89abab21
LLR: -2.92 (-2.94,2.94) <0.00,2.00>
Total: 13184 W: 3285 L: 3535 D: 6364 Elo -6.59
Ptnml(0-2): 85, 1662, 3334, 1440, 71

Failed 180+1.8 VLTC:
https://tests.stockfishchess.org/tests/ ... c881ea908e
LLR: -2.94 (-2.94,2.94) <0.00,2.00>
Total: 64248 W: 16224 L: 16374 D: 31650 Elo -0.81
Ptnml(0-2): 26, 6788, 18640, 6650, 20
=============================================================================
Passed with 8 threads (just a small gain) ->
Passed 60+0.6 th 8 VLTC SMP (STC bounds):
https://tests.stockfishchess.org/tests/ ... 74c2e541dc
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 90630 W: 23372 L: 23033 D: 44225 Elo +1.30
Ptnml(0-2): 13, 8490, 27968, 8833, 11

Passed 60+0.6 th 8 VLTC SMP:
https://tests.stockfishchess.org/tests/ ... da89abacdb
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 137804 W: 35764 L: 35276 D: 66764 Elo +1.23
Ptnml(0-2): 31, 13006, 42326, 13522, 17
With networks and patches like this, soon everyone will need a supercomputer to use Stockfish.
See ya in couple days on Outskirts, need to go now.

cubail · Post by **cubail** » Mon Sep 25, 2023 1:56 pm

yes ıam 3000 kns 8 core

all games draw play chess engıne room

128 core 80.000kns
If ctg books or thousand books are open, there is no loss, all is a draw.

Many people played without books and did not lose

Adrian Oliva · Post by **Adrian Oliva** » Mon Sep 25, 2023 6:16 pm

Skynet wrote: ↑Sun Sep 24, 2023 11:01 pm SF-PB-250923 "Explicitly invoke shell".
https://pixeldrain.com/u/rFwt1Rd*
===============================
Note: Adrian, i wouldn't recommend you to use this version for book tests, because due to larger network on slower PC it gives a big regression, failed on single core ->
Failed 10+0.1 STC:
https://tests.stockfishchess.org/tests/ ... da89abab21
LLR: -2.92 (-2.94,2.94) <0.00,2.00>
Total: 13184 W: 3285 L: 3535 D: 6364 Elo -6.59
Ptnml(0-2): 85, 1662, 3334, 1440, 71

Failed 180+1.8 VLTC:
https://tests.stockfishchess.org/tests/ ... c881ea908e
LLR: -2.94 (-2.94,2.94) <0.00,2.00>
Total: 64248 W: 16224 L: 16374 D: 31650 Elo -0.81
Ptnml(0-2): 26, 6788, 18640, 6650, 20
=============================================================================
Passed with 8 threads (just a small gain) ->
Passed 60+0.6 th 8 VLTC SMP (STC bounds):
https://tests.stockfishchess.org/tests/ ... 74c2e541dc
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 90630 W: 23372 L: 23033 D: 44225 Elo +1.30
Ptnml(0-2): 13, 8490, 27968, 8833, 11

Passed 60+0.6 th 8 VLTC SMP:
https://tests.stockfishchess.org/tests/ ... da89abacdb
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 137804 W: 35764 L: 35276 D: 66764 Elo +1.23
Ptnml(0-2): 31, 13006, 42326, 13522, 17
With networks and patches like this, soon everyone will need a supercomputer to use Stockfish.
See ya in couple days on Outskirts, need to go now.

Hi Alex! Thanks for the update and your recommendation for book tests. I am using 2 threads and do you think I can use this version in engine tests?
Greetings

Skynet · Post by **Skynet** » Mon Sep 25, 2023 11:07 pm

Adrian Oliva wrote: ↑Mon Sep 25, 2023 6:16 pm Hi Alex! Thanks for the update and your recommendation for book tests. I am using 2 threads and do you think I can use this version in engine tests?
Greetings

Hi Adrian, if you want to test it against other engines make sure the opponents engines will have same patches and size of the network.

Adrian Oliva · Post by **Adrian Oliva** » Tue Sep 26, 2023 1:20 am

Thanks Alex! One more thing, the download link does not work, could you reupload it? Thank you so much. Greetings

Skynet · Post by **Skynet** » Tue Sep 26, 2023 8:00 pm

Adrian Oliva wrote: ↑Tue Sep 26, 2023 1:20 am Thanks Alex! One more thing, the download link does not work, could you reupload it? Thank you so much. Greetings

The link is working, corrected.

Skynet · Post by **Skynet** » Mon Oct 02, 2023 7:18 pm

SF-PB-290923_x64 "Update default net to nn-0000000000a0.nnue".

https://pixeldrain.com/u/9ujHss4w

Chess Forum

SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)

Re: SF PB (Stockfish)