[have_faith]

> this changes everything

> As you can see, M1 Mac beats Intel Mac in every Lightroom benchmark across the board

Yeah, by a relatively small amount according to your own charts... It's still impressive, not exactly earth shattering.

[gspr]

… while consuming far less power, while being passively cooled, and while not even running native code!

I think Apple is a huge threat to general purpose computing freedom, and I refuse to use their products. But hats off to their M1!

[ubermonkey]

"I think Apple is a huge threat to general purpose computing freedom"

I don't see how, unless you see literally all proprietary operating systems as such.

Remember, iOS <> MacOS.

[outoftheabyss]

Running through emulation and on an entry level first generation machine that is significantly cheaper

[neogodless]

The Rosetta 2 translation is very impressive.

The $2600 32GB 2TB Intel-based Macbook Pro 13" is only $300 more than the $2300 16GB 2TB M1-based Macbook Pro 13", though. If you could spec the M1 with 32GB, it would probably cost the same.

As pointed out, if you go with 1TB, you save $400. As per usual, the upgrade prices are (in my opinion) crazy!

[ClumsyPilot]

This statement is vacuous unless you can quantify how much faster non-emulated software will be

[temp]

I wouldn't call a $2,300 device "entry level".

[cmiller1]

It's a $1300 entry level device with $1000 worth of maxing ram and SSD space.

[akmarinov]

It is the entry level laptop in the Apple ecosystem, though.

[Guthur]

I've mentioned this a few times but it does seem to be completely glossed over. Intel has for the most part been stuck on 6 year old 14nm and it's positively ancient compared to TSMC 5nm upon which the M1 is manufactured, it's to all intends and purposes two generations ahead. This is big card Apple got to play.

Intel have managed to squeeze an incredible amount of performance out of their 14nm (+++...) but it's at a cost and that cost was power consumption, they had no other choice with their DOA 10nm (hyperbole for affect :)).

[akmarinov]

True, though Intel isn't looking like it's going to catch up in the next 5 years, so it seems that it'll be generations behind for the foreseeable future.

Basically the reason why Apple abandoned them.

[neogodless]

I'm pretty much anti-Intel because I lost 10 years of my life waiting for CPUs to advance (hyperbole!), and it took a swift kick from AMD to get any traction there.

But I also think it's a little short-sighted to predict the next 5 years based on the past 5 years. Intel's Tiger Lake is remarkably good, albeit limited by production capacity (tops out at 4 cores and not found in many products). As an indication of what Intel is capable of, if they start producing CPUs on better processes (before your 5 year prediction), they are very likely to regain the status of at least competitive.

[akmarinov]

I'm not basing the prediction on the past 5 years.

Intel themselves have said that they'll do 7nm around 2022-2023 - https://www.tomshardware.com/news/intel-announces-delay-to-7...

5nm is probably going to be 2024-2025 at this cadence.

And true, Intel's 7nm process might be somewhat equivalent to TSMC's 5nm process as far as performance goes, but at that point in 2022 TSMC will be doing 3nm, according to them.

https://www.imore.com/tsmc-lands-first-3nm-contract-apple-co...

So at the start of 2023, Intel might be at 7nm where TSMC will be at 3nm.

[helmholtz]

I've always wondered how people on HN are able to spit out detailed answers like this with references and facts with such great ease as part of a back-and-forth conversation. Are you a professional in hardware? Or just an enthusiastic follower?

[Guthur]

The various node steps are pretty well defined along with the players.

* For CPU design it's AMD, Intel and ARM players (Apple, Samsung etc)

* For discrete GPU it's really AMD and Nvidia, Intel may be moving into discrete but have traditionally only delivered on within an integrated CPU/GPU package.

* For manufacturing it's Intel, Samsung, TSMC, Global Foundaries.

* And major node steppings at the moment are 14nm, 10nm, 7nm, 5nm, 3nm (there are some less "standard" steppings such as 12nm, 8nm etc. but it's worth noting that these are somewhat marketing monikers at this stage anyway).

And so to find references you can pretty much combine these aspects to come up with the various articles.

I personally love nerding out on chip technology, no major expert just find it interesting to follow.