New Apple M1 CPU

Foxbat

Addicted to new HW
Original poster
Supporting Founder
Pub Member / Supporter
Lifetime Supporter
Nov 25, 2003
22,009
16,321
Michiana
Apple’s Macintosh computers are leaving Intel’s CPUs for an in-house ARM-based M1 CPU. 4 cores are full-powered, 4 are optimized for power-saving activities, and the integrated GPU is another in-house design.

Promoted as 3x the CPU performance of the current Intel Macs and 5x the Intel GPU performance while providing either 2x battery life or 60% lower power consumption for the M1 Mac Mini.

Prices for the new Macs (MacBook Air, 13” MacBook Pro, and Mac Mini) are the same or lower than the current Intel-powered Macs that are these new Macs replace.

 
  • Like
Reactions: dfergie
Bought a new Air earlier this year not knowing these were on the way. Won't need a replacement for a while.

Possibly when the next iMac comes out I will replace my 2013 model.
 
Right now, one limiting factor is the maximum of 16 GB of RAM. I wonder if that’s a limitation of the current A-series of chips (who needs more than 16 GB of storage in a phone/tablet) or if this was just to get the current M1 out in the market now with more Address Lines coming in bigger packages?

It’s not too big a problem for the MacBook Air, but for the MB Pro or the Mini it could be limiting to many potential Pro/Mini buyers.
 
I've been pondering a Mac Mini for some time now as a secondary machine. I'd held out for the new chip versions. It fits the bill perfectly for me. I don't need anything hugely powerful as my PC will still be my primary machine. Throw in Parallels Desktop for windows apps (once it is updated) and add the new capability to run iOS apps natively and this becomes a really good machine that checks a lot of boxes.
 
Promoted as 3x the CPU performance of the current Intel Macs and 5x the Intel GPU performance while providing either 2x battery life or 60% lower power consumption for the M1 Mac Mini.
Yet with all that horsepower, MacOS 11 running on these machines is said to be only twice as fast overall.

Apple talks a big talk about raw power but they seem to have some difficulty bringing it to bear.
 
  • Like
Reactions: Comptech
Yet with all that horsepower, MacOS 11 running on these machines is said to be only twice as fast overall.

Apple talks a big talk about raw power but they seem to have some difficulty bringing it to bear.
That’s one of the reasons I’m waiting to see what the reviews with production machines will reveal.

One thing that I have heard discussed but not verified is the physical location where the M1 Macs’ memory resides. Everyone seems to assume that “SOC” means everything is on the chip, including the System’s DRAM. Apple says the M1 has 16 billion transistors. Excuse my math, but 8 GB of DRAM requires at least 64 billion transistors, four times the published amount.

I believe that the “DRAM” shown in the M1 block diagram of the Unified Architecture is really Level II Cache or the DRAM Memory controllers which could incorporate Level III cache. But not the 8 GB or 16 GB listed in the configurations available for the first pass of M1 Macs.
 
  • Like
Reactions: KE4EST
Looking at this picture, it appears that the DRAM is on the same chip complex, but not on the same die:

chip__fffqz3ljssi2_large_2x.jpg


M1 unifies its high‑bandwidth, low‑latency memory into a single pool within a custom package.
 
  • Like
Reactions: navychop and Foxbat
I think once these actually get out into the real world usage we will know more about how well they perform. But preliminary results seem to be really good. I'm considering the Mac Mini version, but going to wait and see for at least a few months to see how they get rated after they have been in use for a while.

I'm still using my trusty old Mac mini (Mid 2010, 2.4 GHz Intel Core 2 Duo), "tricked" to run macOS Catalina (thanks to dosdude1). Afterwards, I maxed out the memory at 16GB, and upgraded the storage to a 1TB SSD drive, and replaced the dvd burner with a blu-ray burner. The upgrades were not necessary, but added them because I could. :cool:
 
  • Like
Reactions: Foxbat
That makes a lot more sense than having the two memory options tied to two separate versions of the M1. Here, they make all the M1 chips they can, Bin the ones with 8 good GPUs from those with only 7 (MacBook Air entry-level), and marry them to the chip carrier.

Some other clarification from reports, the PCIe and USB may be Version 4, but the Thunderbolt is still Thunderbolt 3 (40 Gbps). Also, there is currently no support for an External GPU via PCIe expansion case.

While I was installing Microsoft Security Updates today, I was reading a news item that said Microsoft is working on a beta version of the Office 365 suite running natively on the Apple Silicon.
 
Running a native ARM benchmark probably isn't a great indicator of how the M1 is going to run software that isn't ARM compiled. Apple has bragged about FCP and one other application that have been ported but I wonder how the performance of applications that haven't been ported are going to do as that may represent the bulk of what end users are running for a while going forward.
 
Once the translation process is complete for each app, I suspect performance will be decent. I have helped some folks in CS who do research on this kind of thing. According to them, going from CISC to RISC is a lot easier than the other way around, like Apple did last time with PPC->x86.
 
I long ago grew tired of the "compared to the previous generation chip. Tell us how this product compares in the real world because that's what we're really interested in at the end of the day.
 
  • Like
Reactions: Foxbat
I find the Integrated Memory interesting in a “what’s old is new” way. It makes perfect sense with Apple taking advantage of the GPU to accelerate certain operations, but I’m reminded of the days of graphics cards that used up System RAM, so that 8 MB you thought you could use turned into 6 MB because the graphics grabbed the other 2MB.

Here, macOS and the M1 are saying data and instructions are equally important and accessible. The program calls on something to happen, and it doesn’t need to wait while the buffer transfers through the PCIe bus to the GPU to be processed, then transfer back again to the next stage.

But unlike the Mac Pro, these M1 Macs will be more affordable to Apple users. I imagine that Apple will sell more MacBook Airs in this first wave, mostly because of the sub-$1000 price (by a whole dollar, I know).

The Mac mini might make sense for someone looking into a Mac for the first time, or a current Mac mini owner with a 2014 or earlier model.

My speculation is that Apple could make a Mac nano about the size of the AppleTV with reduced core count and 4 or 8 GB, and sell it for $399. That could be interesting…
 
It makes perfect sense with Apple taking advantage of the GPU to accelerate certain operations, but I’m reminded of the days of graphics cards that used up System RAM, so that 8 MB you thought you could use turned into 6 MB because the graphics grabbed the other 2MB.
I hadn't thought about the video RAM being shared with the system RAM but since it is all on the same chip, you really don't have any choice. 16GB of RAM may be getting awfully tight with Apples obtuse display sizes and however much RAM is required by the AI modeling.
 
I’m considering replacing my 2016 MacBook Pro 13” Escape-Key edition (aka non-TouchBar) with the new MBA. I was never happy with its battery life compared to my old MacBook Air. I can retire this and the iPad Air I use to surf.

Update: Yep, my 2016 MBP is worth $420 in trade. Thinking…:lovecomputer
 
Last edited:

Similar threads

Users Who Are Viewing This Thread (Total: 0, Members: 0, Guests: 0)

Who Read This Thread (Total Members: 1)

Top