M1 Apple Silicon for Chess?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
acepoint_de
Posts: 36
Joined: Mon Jun 10, 2013 11:14 pm

Re: M1 Apple Silicon for Chess?

Post by acepoint_de » Sun Feb 21, 2021 12:27 pm

AlexChess wrote:
Sun Feb 21, 2021 8:17 am
Could you please compile also MikeB's Honey R12? Already Honey XI Intel 64 is doing very well!
Honey-v13 (and Black-Diamond/Oki-Maguro) are finished, uploaded and on the benchmark list: https://acepoint.de/some-chess-stuff-for-apples-m1/

@MikeB Bluefish and Harmon are not compiling using your provided script users_build.sh (I have changed ARCH to apple-silicon and COMP to clang). But I also don't know the differences between the five engines you provide at https://github.com/MichaelB7/Stockfish/ ... s/tag/13.0. As others already asked in the Honey 13 thread ;-).

Ciao

acepoint

Ckappe
Posts: 64
Joined: Sun Feb 14, 2021 10:50 am
Full name: Rütger Andersen

Re: M1 Apple Silicon for Chess?

Post by Ckappe » Sun Feb 21, 2021 1:23 pm

BetaPro wrote:
Sun Feb 21, 2021 4:45 am

Yes 5800U will be better for laptops as 5900H is 45W.

When I was working on graphics drivers for those Cezanne parts. I was assigned an issue where QA could only reproduce on 45W parts. But the dev team had those engineering samples where it's hard to tell if it's a 45W or a 15W. I asked the team, and my teammates also didn't know a surefire way to determine that. So we had to go to the product manager who had a list telling us about the specs of each engineering sample..
Better is subjective :-)

As always its a tradeoff Energy consumption and performance :-) (just look at the Wattage of those 3090 GPUs :-) ) .. Personally I don't need the more extreme battery-stamina running a laptop 3x8h working days on one charge. Running it 4-8 hours normal processing when traveling is just fine between charges for me. I'd rather lose the extreme battery stamina than give up top performance of my laptop. As so many laptops these days have 90-99W batteries, the larger batteries can really give "good enough" stamina and still provide the power to run beefier HW even in laptops. For me I'd rather get more done in a shorter period of time and "recharge" rather than get less work done over a longer time-period.. :-)

For me personally I think 45W TDP 7nm strikes the perfect balance between power-efficiency this spring.. the current 15W options like the i7-10710U (Apple published TDP for M1 is 39W btw which is quite high at full use!!) are a bit too slow to justify the extra energy savings imop.

AlexChess
Posts: 72
Joined: Sat Feb 06, 2021 7:06 am
Full name: Alex Mora

Re: M1 Apple Silicon for Chess?

Post by AlexChess » Sun Feb 21, 2021 4:30 pm

acepoint_de wrote:
Sun Feb 21, 2021 12:27 pm
AlexChess wrote:
Sun Feb 21, 2021 8:17 am
Could you please compile also MikeB's Honey R12? Already Honey XI Intel 64 is doing very well!
Honey-v13 (and Black-Diamond/Oki-Maguro) are finished, uploaded and on the benchmark list: https://acepoint.de/some-chess-stuff-for-apples-m1/

@MikeB Bluefish and Harmon are not compiling using your provided script users_build.sh (I have changed ARCH to apple-silicon and COMP to clang). But I also don't know the differences between the five engines you provide at https://github.com/MichaelB7/Stockfish/ ... s/tag/13.0. As others already asked in the Honey 13 thread ;-).

Ciao

acepoint
Thank you acepoint and thank you MikeB for the engines. I have made a nice OKI-Maguro Logo with Japanese food and a Samurai chessboard. https://1drv.ms/u/s!AkW3Hj0Gl_ewzxWooPL ... c?e=ybtcbV
For now I have emulated Fat Fritz 2.0 using Stockfish 12 ARM64 and the FatFritz2.v1.bin in the Stockfish FF GitHub folder, and it is probably stronger than the original-clone :lol:

AlexChess
Posts: 72
Joined: Sat Feb 06, 2021 7:06 am
Full name: Alex Mora

Re: M1 Apple Silicon for Chess?

Post by AlexChess » Mon Feb 22, 2021 5:37 am

@MikeB @Acepoint_de
Oki-Maguro v13, Honey v13 and Black Diamond v13 on Silicon M1 are strongs!
Stockfisk 12 doesn't work with FatFritz2v1.bin (I haven't tested the engine before starting the BanksiaGui blitz tournament)

AlexChess
Posts: 72
Joined: Sat Feb 06, 2021 7:06 am
Full name: Alex Mora

Re: M1 Apple Silicon for Chess?

Post by AlexChess » Mon Feb 22, 2021 10:32 am

Just completed my superblitz tournament on Mac mini M1
Last edited by AlexChess on Mon Feb 22, 2021 10:38 am, edited 1 time in total.

AlexChess
Posts: 72
Joined: Sat Feb 06, 2021 7:06 am
Full name: Alex Mora

Re: M1 Apple Silicon for Chess?

Post by AlexChess » Mon Feb 22, 2021 10:35 am

@ACEPOINT_DE @MikeB OKI MAGURO beyond Stockfish 13!!!!

Blitz 3 min + 3 sec BanksiaGUI M1 3 up to 12 Millions nodes / second NO BOOK 6-man sygyzy Tablebases
Fat fritz 2.0 (emulated using Stockfish 12 + FatFritz2v1.bin) NOT WORKING

Result:
-------------------------------------------------------------------------------------------
# name games wins draws losses score elo + -
1. SugaR AI 1.50 17 11 6 0 14.0 221 147 128
2. Cfish 170221 64 NEON 17 10 6 1 13.0 179 140 126
3. Oki Maguro v13 17 8 9 0 12.5 156 128 118
4. Stockfish 13 17 7 10 0 12.0 135 125 117
5. Dragon by Komodo Chess 64-bit 17 7 9 1 11.5 128 126 119
6. Honey v13 17 7 9 1 11.5 118 126 120
7. Black Diamond v13 17 7 9 1 11.5 137 126 119
8. Crystal 170221 17 6 10 1 11.0 119 123 118
9. ShashChess Pro 2.0 64 POPCNT 17 6 8 3 10.0 54 127 125
10. Komodo 14.1 64-bit 17 5 8 4 9.0 49 123 123
11. Ethereal 12.75 (POPCNT) 17 5 6 6 8.0 -11 126 128
12. Xiphos 0.6 SSE 17 4 6 7 7.0 -55 125 131
13. Lc0 v0.26.3+git.dirty 17 2 7 8 5.5 -101 125 135
14. RubiChess 1.6.1.1 17 2 6 9 5.0 -123 126 139
15. Texel 1.08a13 17 2 5 10 4.5 -149 129 145
16. Igel 2.9.0 CUSTOM 17 3 2 12 4.0 -173 137 162
17. Marvin 5.0.0 17 1 4 12 3.0 -214 136 162
18. Fat Fritz 2.0 GITHUB 17 0 0 17 0.0 -469 185 326

Cross table:
-------------------------------------------------------------------------------------------
# name score games 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1. SugaR AI 1.50 14.0 17 x 0.5 0.5 0.5 1.0 1.0 0.5 0.5 1.0 0.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
2. Cfish 170221 64 NEON 13.0 17 0.5 x 0.5 0.0 0.5 0.5 1.0 0.5 0.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
3. Oki Maguro v13 12.5 17 0.5 0.5 x 0.5 0.5 0.5 0.5 0.5 0.5 0.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
4. Stockfish 13 12.0 17 0.5 1.0 0.5 x 0.5 0.5 0.5 0.5 0.5 0.5 0.5 1.0 1.0 0.5 1.0 1.0 1.0 1.0
5. Dragon by Komodo Chess 64-bit 11.5 17 0.0 0.5 0.5 0.5 x 0.5 0.5 0.5 1.0 0.5 1.0 1.0 1.0 0.5 1.0 0.5 1.0 1.0
6. Honey v13 11.5 17 0.0 0.5 0.5 0.5 0.5 x 0.5 0.5 1.0 1.0 0.5 1.0 1.0 1.0 0.5 0.5 1.0 1.0
7. Black Diamond v13 11.5 17 0.5 0.0 0.5 0.5 0.5 0.5 x 0.5 0.5 0.5 1.0 0.5 1.0 1.0 1.0 1.0 1.0 1.0
8. Crystal 170221 11.0 17 0.5 0.5 0.5 0.5 0.5 0.5 0.5 x 0.5 1.0 0.0 0.5 1.0 0.5 1.0 1.0 1.0 1.0
9. ShashChess Pro 2.0 64 POPCNT 10.0 17 0.0 0.5 0.5 0.5 0.0 0.0 0.5 0.5 x 1.0 1.0 1.0 0.5 1.0 0.5 1.0 0.5 1.0
10. Komodo 14.1 64-bit 9.0 17 0.5 0.0 0.5 0.5 0.5 0.0 0.5 0.0 0.0 x 0.5 0.5 1.0 0.5 1.0 1.0 1.0 1.0
11. Ethereal 12.75 (POPCNT) 8.0 17 0.0 0.0 0.0 0.5 0.0 0.5 0.0 1.0 0.0 0.5 x 0.5 1.0 0.5 0.5 1.0 1.0 1.0
12. Xiphos 0.6 SSE 7.0 17 0.0 0.0 0.0 0.0 0.0 0.0 0.5 0.5 0.0 0.5 0.5 x 1.0 0.5 1.0 0.5 1.0 1.0
13. Igel 2.9.0 CUSTOM 4.0 17 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.5 0.0 0.0 0.0 x 1.0 0.5 0.0 1.0 1.0
14. Lc0 v0.26.3+git.dirty 5.5 17 0.0 0.0 0.0 0.5 0.5 0.0 0.0 0.5 0.0 0.5 0.5 0.5 0.0 x 0.0 1.0 0.5 1.0
15. RubiChess 1.6.1.1 5.0 17 0.0 0.0 0.0 0.0 0.0 0.5 0.0 0.0 0.5 0.0 0.5 0.0 0.5 1.0 x 0.5 0.5 1.0
16. Texel 1.08a13 4.5 17 0.0 0.0 0.0 0.0 0.5 0.5 0.0 0.0 0.0 0.0 0.0 0.5 1.0 0.0 0.5 x 0.5 1.0
17. Marvin 5.0.0 3.0 17 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.5 0.0 0.0 0.0 0.0 0.5 0.5 0.5 x 1.0
18. Fat Fritz 2.0 GITHUB 0.0 17 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 x

Tech:
-------------------------------------------------------------------------------------------

Tech (average nodes, depths, time/m per move, others per game), counted for computing moves only, ignored moves with zero nodes:
# name nodes/m NPS depth/m time/m moves time #fails
1. SugaR AI 1.50 26323K 7281712 34.4 3.6 50.8 183.7
2. Cfish 170221 64 NEON 31276K 8826226 35.6 3.5 58.4 207.1
3. Oki Maguro v13 28042K 7586301 36.0 3.7 59.2 218.8
4. Stockfish 13 23435K 6501897 35.3 3.6 57.8 208.2
5. Dragon by Komodo Chess 64-bit 18768K 4349553 27.7 4.3 53.1 229.3
6. Honey v13 4319K 1174957 23.7 3.7 62.9 231.3
7. Black Diamond v13 27045K 6953408 26.4 3.9 61.4 238.7
8. Crystal 170221 22890K 6130108 25.1 3.7 59.2 221.0
9. ShashChess Pro 2.0 64 POPCNT 35292K 9389806 27.9 3.8 58.3 219.2
10. Komodo 14.1 64-bit 44503K 10523284 30.3 4.2 63.8 269.8
11. Ethereal 12.75 (POPCNT) 45452K 11932348 28.5 3.8 67.0 255.2
12. Xiphos 0.6 SSE 39004K 10659102 29.6 3.7 68.9 252.1
13. Igel 2.9.0 CUSTOM 23640K 6505623 22.9 3.6 54.9 199.4
14. Lc0 v0.26.3+git.dirty 31K 8265 7.7 3.8 56.1 211.9
15. RubiChess 1.6.1.1 31463K 8540386 29.5 3.7 63.2 232.8
16. Texel 1.08a13 34055K 8787894 20.8 3.9 59.9 232.3
17. Marvin 5.0.0 14630K 3571664 20.3 4.1 46.4 189.9
18. Fat Fritz 2.0 GITHUB 0K 0 0.0 0.0 0.0 0.0 17
all --- 26303K 7104334 27.3 3.8 58.1 220.1 17

AlexChess
Posts: 72
Joined: Sat Feb 06, 2021 7:06 am
Full name: Alex Mora

Re: M1 Apple Silicon for Chess?

Post by AlexChess » Tue Feb 23, 2021 1:10 pm

AlexChess wrote:
Sun Feb 21, 2021 4:30 pm

Could you please compile also MikeB's Honey R12? Already Honey XI Intel 64 is doing very well!
Honey-v13 (and Black-Diamond/Oki-Maguro) are finished, uploaded and on the benchmark list: https://acepoint.de/some-chess-stuff-for-apples-m1/

@MikeB Bluefish and Harmon are not compiling using your provided script users_build.sh (I have changed ARCH to apple-silicon and COMP to clang). But I also don't know the differences between the five engines you provide at https://github.com/MichaelB7/Stockfish/ ... s/tag/13.0. As others already asked in the Honey 13 thread ;-).

Ciao

acepoint



Thank you also for RubiChess 2.1 NNE and Lc0 0.27.0 ...but SF Fat Fritz 2.0 Github for M1? I have read that someone already compiled it for Windows 10 under GPL 3.0. It would be interesting to test it against ore strong Stockfish derivates like Oki-Maguro v13 (Thanks again MikeB)) I'm trying to learn how to compile it for myself, but as a .net web programmer I have never done it before... :wink:

Stockfish GPL 3.0 Fat Friitz 2.0 github repository : https://github.com/DanielUranga/Stockfi ... 3426c93c6e

acepoint_de
Posts: 36
Joined: Mon Jun 10, 2013 11:14 pm

Re: M1 Apple Silicon for Chess?

Post by acepoint_de » Tue Feb 23, 2021 1:32 pm

AlexChess wrote:
Tue Feb 23, 2021 1:10 pm
Thank you also for RubiChess 2.1 NNE and Lc0 0.27.0 ...but SF Fat Fritz 2.0 Github for M1? I have read that someone already compiled it for Windows 10 under GPL 3.0. It would be interesting to test it against ore strong Stockfish derivates like Oki-Maguro v13 (Thanks again MikeB)) I'm trying to learn how to compile it for myself, but as a .net web programmer I have never done it before... :wink:

Stockfish GPL 3.0 Fat Friitz 2.0 github repository : https://github.com/DanielUranga/Stockfi ... 3426c93c6e
I know where it is. But I'm having trouble with compiling this case of plagiarism and putting it online ;-).

Ciao

acepoint

AlexChess
Posts: 72
Joined: Sat Feb 06, 2021 7:06 am
Full name: Alex Mora

Re: M1 Apple Silicon for Chess?

Post by AlexChess » Tue Feb 23, 2021 3:55 pm

OK,
so if also you have problems, I can stop trying and just wait :oops:

cpeters
Posts: 15
Joined: Wed Feb 17, 2021 6:44 pm
Full name: Christian Petersen

Re: M1 Apple Silicon for Chess?

Post by cpeters » Tue Feb 23, 2021 4:00 pm

I'm reallyreally tempted to pm him a link to a silicon*-compiled version with the AS-net, just to do a little piracy/HARMLESS FUN here.

Image: dropping a "Fritz"-engine for a platform not even officially supported before they do. Unheard of!

l0lz0rs


*I'll wait for the next gen of the chips, and then buy me a fine mini!

Post Reply