M1 Apple Silicon for Chess?

[syzygy]

I find all these negative reactions interesting. In my view, Apple has shown that x86 is going the way of the light bulb.

[Jouni]

Apple claims by far the best performance CPU/GPU for watt. Even with 10W a lot of speed! But can You run it 24/7?

[Ras]

In my view, Apple has shown that x86 is going the way of the light bulb.

ARM vs x86 is not the only factor here. It's also that Apple is using TSMC's 5nm production process while even the best x86 CPUs (i.e. AMD's) are on 7nm. Meanwhile, Intel is struggling to get their 10nm process (comparable to TSMC's 7nm) right so that Intel is actually two production process generations behind.

Comparisons where you change two variables and then attribute the resulting differences to only one of them are a logical mistake.

[Dann Corbit]

Something powerful and efficient from apple is very good news.
First of all, the apple users have been shortchanged recently, with dollars per mip making a poor showing on apple hardware.
This new stuff seems pretty good, especially for laptops.
And IBM was not pushing AMD at all. Maybe this sort of thing will make them play the 5nm card earlier than they wanted to.
Competition like this is good to push the industry forward.
Personally, I don't care who makes the chip. I just want chess analysis to pour out of it.

[syzygy]

Apple claims by far the best performance CPU/GPU for watt. Even with 10W a lot of speed! But can You run it 24/7?

Of course you can. Since it doesn't consume much power, it doesn't produce much heat and it is easy to keep cool. M1 has just 4 fast cores but is only the beginning.

[Ras]

Apple claims by far the best performance CPU/GPU for watt. Even with 10W a lot of speed! But can You run it 24/7?

You can, but since the Macbook Air M1 has no fan, it will enter thermal throttling after about 10 minutes of sustained load, leading to a performance loss of about 15%. The Macbook Pro M1 does have a fan that kicks in on sustained load (but still very quiet) and thus won't throttle.

[syzygy]

Comparisons where you change two variables and then attribute the resulting differences to only one of them are a logical mistake.

Who are you talking to exactly? I did not even give you my reasoning, only the conclusion.

[Ras]

I did not even give you my reasoning, only the conclusion.

Yeah, from Apple presenting an SoC that cannot be compared to x86 CPUs in the first place because of 5nm production and using LPDDR4-4266. Only that both of them will be, or even are (depending on the OEM) open to x86, too.

[syzygy]

I did not even give you my reasoning, only the conclusion.

Yeah, from Apple presenting an SoC that cannot be compared to x86 CPUs in the first place because of 5nm production and using LPDDR4-4266. Only that both of them will be, or even are (depending on the OEM) open to x86, too.

?Syntax error

[MRI_Doc]

You get max performance with 4 threads, which utilizes only the high performance cores.

Adding more threads results in SLOWER calculation, as the code must not be optimized for variable speed cores.

Run your benchmarks with 4 threads and the M1 performs very well.