Apple is an existential threat to the PC

No, they're silly because they run contradictory to the results and verdicts of other reputable -- and less gaming/Adobe focused -- reviews.

Also, I really don't appreciate you claiming to know what I do and do not want to see.

Efficiency = performance/inputs.

I would hope so at more than double the power consumption.

What?
The MSI review units sent out cost $4000.

 

Also, it occurred to me just how inconsistent the argument you're putting forward is:

TSMC 5nm vs Intel 7 isn't enough to explain a ~2.5x difference in power consumption.

If you're fine with the vast difference in power consumption between the GE76 and the MBP, I assume you would be fine in comparing whatever Apple Silicon chip Apple puts in their upcoming Mac Pro to ADL mobile? Why stop there -- let's just compare overlocked CPUs to stock CPUs, EPYC to desktop ADL. Who cares about power consumption, right?

Thank God for sub-forums then!

 

Not me
I wouldnt get an intel though, CPUs sounding like vacuum cleaners just arent my thing, My last intel chip I had to downclock 20% and even then it made a shitload of noise

Apple MacBook Pro 16 2021 M1 Max Witcher 3 ultra *39.5db
MSI GE76 Raider i9-12900HK, GeForce RTX 3080 Ti Laptop GPU Witcher 3 ultra *53db

though I assume the 3080 performs better.
>50db a meter away would annoy the shit out of me

 

Should probably also remind everyone that they are comparing the old cores. The Avalanche and Blizzard cores in A15 are more efficient while also delivering better performance on the same, albeit optimised node. SpecCPU 2017 single core shows a 2.5% to 37% improvement compared to the high performance cores used in M1 / Pro / Max. The energy efficient cores show up to an 46% performance uplift. That's in the iPhone form factor as well and much less memory bandwidth.

 

Intels next process node and CPU will see improvements, too.

 

Good point.

Speaking of which:
https://twitter.com/DSCCRoss?ref_src=twsrc^tfw|twcamp^tweetembed|twterm^1488224586290237440|twgr^|twcon^s2_&ref_url=https://www.macrumors.com/2022/01/31/imac-pro-2022-summer-launch/

I think this mini-LED equipped iMac Pro will probably use M1 Pro/Max. There might be a higher TDP SKU (labelled as something like M1 Ultra or M1Z Max -- oh, maybe M1 Maxx! ) with two M1 Max dies on the same package. I suspect an Apple Silicon equipped Mac Pro will use a completely different configuration to the MBP and iMac Pro. Something that will be built with a stupid amount of PCI-E lanes. But I am still completely baffled as to how they will manage AMD GPUs. Unlike most, I don't think Apple will drop support for MPX modules and AMD Radeon Pro GPUs.

 

Well the thread title is literally about Apple threatening the PC. So I guess it's dedicated to a VS discussion?

 

Hopefully!

But to suggest they will suddenly bridge the performance/watt disadvantage within the next generation is a bit of a stretch.

I’m really intrigued to see what Apple could do in a desktop environment though.

 

Intel won't be competitive in perf/power until they get EUV into mass production, which should happen in the latter half of this year, - 3 years after TSMC started using EUV for their N7+ in the bottom 4 metal layers.

I will never cease to amaze me how Intel went from being 18 months ahead of everyone else in process tech to being three years behind, - in 7 years.

Cheers

 

Shows the importance of process nodes.

 

Yeah, I'm curious to see what the extra TDP headroom actually provides on this platform. We all know it's going to go faster, the question will be how much is the M1/M2 optimized for purely low power operating environment vs outright speed? Or said another way, how hard will it tank the power efficiency when a bit more wattage is pumped through it to pick up some extra oomph? Everyone knows there's a sliding scale here, and the top-end of that scale asymptotes at infinite power consumption. Wonder how far they move through the J-portion of the curve trying to squeeze more performance?

Precisely where my head is too. It's nuts how far they fell in such a short time.

 

What's preventing any other company rolling their own OS (macOS is a unix-derivative), licensing ARMs IP and modifying it for their OS, or developing their own GPU cores? You don't have to acquire whole companies to get expertise.

Most of Apple's technology is simply Apple-flavoured versions of technology already out there.

Microsoft should easily be able to do something akin to Apple's ARM transition.

 

Apple has always been a hardware company since the beginning. The only reason ARM exist in its current form is because of Apple's work with the Newton in the 1990s.

If it's so easy why aren't there more doing it? All the big ones have tried and fell short, be it Microsoft, Intel, Google or Amazon. It's a massive undertaking with expensive engineers and supply chain nightmares.

 

That's mental. Was Apple party to the decision for Acorn to spin its evolving RISC architecture into its own company? Yes. Were Apple responsible? No. Apple's interest in RISC/ARM-driven CPUs can be mapped to 1992 (the Newton) then zip until fifteen years later (iPhone). Unless you wish to count the über low-powered ARM microcontrollers included in iPods. That's a stretch given the iPod wasn't a software platform.

In what way have Microsoft, Intel, Google or Amazon tried (and failed) to replicate Apple going all-in on building their own software on their own hardware?

Microsoft have their own OS but have never tried optimising their own OS/software for their custom own CPU/GPU architecture. Intel don't have their own OS. Google's plan for Android was that it wouldn't be tailored for their own specific hardware but that it would run on a wide variety of hardware. Amazon neither have their own OS, nor their own CPU or even their own hardware.

Why hasn't Microsoft invested fully into a vertically integrated system for general computing, whereas they have for games (Xbox)? Who the hell knows. Developing and gaining traction for a new OS looks to me to be the biggest barrier. Taking the ARM IP and tailoring that for your OS, software and hardware looks like the least amount of effort. Apple have chopped-and-changed CPU architectures three times now (680x0, PowerPC, Intel, ARM). Done right, it doesn't present any kind of barrier.

 

Someone with sense, couldnt agree more.

 

I think the overall answer we are looking for is just that any of the above isn't easy. As you mentioned all the big companies just doesn't, currently, have the expertise to do it. It would take many years of effort to reach the point where Apple is now and cost billions in investments.

Hopefully there will be an event the 8th of March with a reveal of the M2 chip.

 

The biggest barrier to a fully vertically-integrated ecosystem is an OS that people want to use. On the desktop, this gives you Windows, macOS, linux and ChromeOS and only one of these (macOS) is heavily-optimised for the hardware. linux is designed to run on a wide variety of hardware, where optimisation would need extra work and ChromeOS is architected to abstract everything anyway. Microsoft could optimise Windows better for specific hardware but that's not really their current Windows engineering philosophy, outside specific niches where you get optimised subsets of Windows functionality in Xbox.

The next barrier is the hardware. This is the only Microsoft's barrier to rolling their own PC with their CPU/GPU and a version of Windows that is compiled to be very optimised for that particular version of Windows. This is all Apple do, they don't have fancy unique ARM dev tools, it's just LLVM. Licensing the ARM IP and improving it to run your software stack is what almost every ARM licensee does. Knowing what you need to change to improve performance does rely on understanding the utilisation of your OS and software of the hardware so it's kind of fortunate that Microsoft have decades of experience producing development kits and optimisation tools that would make this much easier.

Microsoft have chosen not too. I don't buy for a minute that this is complex or beyond Microsoft's technical ability in any way. ARM's CPU and GPU architecture are open and well-documented. and big and small companies alike are taking those reference deigns and making them work better for the specific applications that they are required for - and this Wikipedia article lists each major ARM architecture, along with all of the variations of that base architecture that other companies have taken and made unique versions of, along with some of the products that these things have ended up in.