It's not a fair question. What’s the point of ARM? is at best a stupid question. The point of ARM is to be a RISC processor and the point of RISC is much higher efficiency. Energy as well as DIE space efficiency. And the point of efficiency is not just reduced heat and energy consumption, but also to enable much higher performance even for desktops and high performance computing. ARM is by no means reduced to being used as a phone or laptop processor. That's just a stupid assumption.
Any link to support that Mx SoCs are more die space efficient?
No link, but that's the whole point of reducing the overblown CISC instruction set. You don't need to implement the logic for the more esoteric instructions nobody ever needs, therefore an ARM core needs fewer transistors, consumes less energy and produces less heat. And it's also smaller on the DIE. Every 5nm transistor still consumes the same amount of energy, but RISC processors need much fewer of them to build a whole CPU core.
The AMD non-X CPUs may say 65W on the tin but that number is meaningless. AMD may as well write 45W, or 72W, and it would be just as useful because the real power consumption is closer to 90W (AMD Power Consumption).
You forgot one important part, the simplicity should allow for higher frequencies. You are also stuck in the past.
What's Jesus' birthday got to do with it? Did the law of physics change since the 90s?
But the M1 Ultra includes 64 GPU cores and 32 Neural Engine cores! If we assume that these chips are produced on the same process nodes, than the energy consumption differences are direct proportional to the number of transistors used for a calculation.
The relative speed between CPU and RAM has changed since the 90s, the available transistor budget has changed since the 90s, the design tools have changed since the 90s, and we've gained a heck of a lot of real-world experience that, in some cases, has taught us that research and theory don't always translate into the real world.
I see. Everything changed when Gari Halliwell left the Spice Girls.
So the laws of physics haven't changed? Good to know. The point of RISC over CISC is still efficiency. Everything else that improved efficiency doesn't change that fact. ARM didn't stop to be a RISC architecture in the 90s. Neither did x86-64 stopped being CISC. It's still all the same for the same reasons.
Do you understand the differences between the classic 5-stage RISC pipeline and a modern super-scalar out-of-order processor as well as the reasons for the differences?
No it's the exact opposite. Apple's special video decoders and image signal processors are designed to do just one kind of calculation with the utmost speed and efficiency. So they have an even smaller instruction set and even fewer transistors involved in its execution. The entire chip has more and more transistors, because of all the many engines specialized on doing only one task. But a single engine, which for example only decodes x264 for video playback, is itself as tiny as it can be. The rest of the chip draws no power, while only one engine works on the task. That's how Apple A-series and M-series chips can have dozens of hours of video playback on battery life. Higher efficiency always means fewer transistors are actively involved.
In Apple's case? The point is to not have to buy another company's chips or pay an x86 licensing fee I'd guess.
Based on rumors it kinda sounds like Apple agrees. It seems like the only reason they're keeping the Mac Pro form factor around is for those delicious PCI-e slots used in an increasingly small segment of workloads. We might not see ARM powered Mac Pros around for very long.
Since Apple uses multiple coprocessors, the Mx SoC tends to be very large. In fact, there must be very few chips larger than the M1 Ultra. One of them is the AMD MI300, a giant APU.
Ha! And best of luck to you!As a Hackintosh user...
...the longer there is an Intel machine in the first party lineup...
