M1 Apple Silicon for Chess?

[jshriver]

Future might be Risc-V regardless but Nvidia-ARM will definitely make risc-v more attractive.

As an architecture geek I'm rooting for Risc-V too, or at least enough to see it have commercial adoption. Would love to have a box now, but the few SBC's I've seen have been pretty cost prohibitive and more for testing purposes.

[Alayan]

So you agree that x86 is going the way of the light bulb.

I take you not bothering to answer to my message as meaning that you have not a single counterpoint to what I said.

[syzygy]

So you agree that x86 is going the way of the light bulb.

I take you not bothering to answer to my message as meaning that you have not a single counterpoint to what I said.

Or I just spend my time wisely. My intention here was just to point out the in my view unwarranted negativity. It reminds me of the initial reception in this forum of Alpha Chess Zero (and many here will probably still deny that LC0 wouldn't have existed without Deepmind's papers).

[AlexChess]

Benchmarks on a MacBook Air 16GB 8-core GPU

Stockfish compiled for M1 Apple Silicon, commit f9595828eb7e5e970b0be3ee5f84ddd726845523 Wed 11 Nov

Code: Select all

bench	2408119
bench 64 1 20	2270034
bench 64 2 20	4503092
bench 64 4 20	9069926
bench 64 6 20	10820178
bench 64 8 20	12438598

Cfish compiled for M1 Apple Silicon, commit 6193ed1c3809cb4b71ad8f630fa4f52160390fb6 Sun 15 Nov

Code: Select all

bench	2844055
bench 64 1 20	2614400
bench 64 2 20	5232224
bench 64 4 20	10152288
bench 64 6 20	12388337
bench 64 8 20	14935531

dragon-ox running under Rosetta-2

Code: Select all

command  nodes  sec      nps
bench 1	1915866	4.80	399139
bench 2	3653649	4.70	777372
bench 4	5690302	3.36	1693542

lc0, compiled for M1 running on CPU using Apple's Accelerate framework, output of benchmark first position only:

Code: Select all

--backend=blas --weights=J92-330	 97
--backend=blas --weights=T70net-703810   1703

lc0, compiled for M1 running on GPU using OpenCL, output of benchmark first position only:

Code: Select all

--backend=opencl --weights=J92-330	   104
--backend=opencl --weights=T70net-703810   3431

Hi! Could you please give me the link to download LC0 and Cfish binary for Silicon M1? I already have Stockfish M1 (1 milion nps versus 6200 knps with Rosetta 2 on my Mac mini M1 8-256 GB) I can report here the results of my engine vs engine matches if you are interested.

Thank you in advance, AlexChess

[kiroje]

I would really like to know how this one would perform https://shop.solid-run.com/product/SRLX ... E064H08CH/

Anyone that would speculate on that ?

[AlexChess]

I would really like to know how this one would perform https://shop.solid-run.com/product/SRLX ... E064H08CH/

Anyone that would speculate on that ?

Me
Yes, HoneyComb is wonderful.
If Trusted Computing wins, it will be my way. Ready to use Linux on ARM & open source. But until now, simply lowering a little the system protection, I'm able to use every application and OS with my surprising fast Mac mini M1. So, returning to the topic, I want to understand how to compile NATIVE Silicon M1 chess engines (I already have Stockfish ARM64 *1.1 Mnps* that easily wins against Ethereal 12.75 64 bit Rosetta 2)

[BetaPro]

Comparison of Stockfish and Cfish running on 8 cores, classical v NNUE, on M1 and iMac Pro Xeon W

run using

stockfish|cfish 64 8 20 default depth classical|NNUE

Code: Select all

		        Stockfish	Cfish
M1	classical	16319718	17118388
M1	NNUE	        10100445	12759621
Xeon W	classical	16900088	17929299
Xeon W	NNUE	        13631405	14191552

iMac Pro compile options:

Code: Select all

george@Georges-iMac-Pro-2 ~ % stockfish 
Stockfish 141120 by the Stockfish developers (see AUTHORS file)
compiler

Compiled by g++ (GNUC) 10.2.0 on Apple
Compilation settings include:  64bit AVX512 BMI2 AVX2 SSE41 SSSE3 SSE2 POPCNT
__VERSION__ macro expands to: 10.2.0

quit
george@Georges-iMac-Pro-2 ~ % cfish
Cfish 141120 64 BMI2 by Syzygy based on Stockfish
compiler

Compiled by gcc (GNUC) 10.2.0 on Apple
Compilation settings include: 64bit AVX512 BMI2 AVX2 SSE41 SSSE3 SSE2 POPCNT
__VERSION__ macro expands to: 10.2.0

quit

From your benchmark, it looks like it's pretty solid, in NNUE it will be slower because ARM lacks good AVX support.

The SF classic result is much better, 4 big core and 4 small core matching Xeon's 8 big cores.

Can you also run just one thread for SF classic, for both M1 and Xeon? I'm very interested in single threaded performance of a CPU.

Thanks!

[BetaPro]

Comparison of Stockfish and Cfish running on 8 cores, classical v NNUE, on M1 and iMac Pro Xeon W

run using

stockfish|cfish 64 8 20 default depth classical|NNUE

Code: Select all

		        Stockfish	Cfish
M1	classical	16319718	17118388
M1	NNUE	        10100445	12759621
Xeon W	classical	16900088	17929299
Xeon W	NNUE	        13631405	14191552

Actually I got impatient so I went on Scaleway and rented an M1.

Here's what I got with Stockfish Classical
./stockfish bench
3568884 nps

./stockfish bench 64 1 20
3024394 nps

./stockfish bench 64 8 20
17084333 nps

The Macbook 2019 that my workplace provided, which has a single thread Geekbench score of around 1050 gets this,
./stockfish bench
2168394 nps

./stockfish bench 64 1 20
1908261 nps

The Intel version was compiled with x86-64-bmi2. So the m1 is about 64% faster in ST than the Intel Mac, about in line with what Geekbench suggests. Would love it if someone with the latest Zen 3 do a similar test (especially single thread).

[Gian-Carlo Pascutto]

Performance of Stockfish dev from today looks like:

Code: Select all

Zen1 1700X 3.2GHz:     1520000
Ivy Bridge 3.5GHz:     1760000
Haswell 3.4GHz:        1860000
M1 3.2GHz:             2390000
Zen2 3900X 3.8-4.2GHz: 2490000

This is on an M1 Air. I love the fact that it's 100% fanless.

[BetaPro]

That's also a pretty good result for M1, matching Zen 2 while having weaker vector instructions.

I'll pick an M1 up eventually. Overall seems better than any alternative. Pretty sure they will provide higher core count models as well, but I dont really need that many cores.