• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Apple M1 CPU

0h_no_not_again said:
That is a really beautiful piece of data. What is striking to me is the better branch prediction of the A12 over the 9900k, which wobbles about but probably accounts for a lot of those ~10% extra cycles.
Click to expand...

Could that be because RISC has less possible branch prediction routes that it can go down, so with less options to choose from, it automatically becomes better at prediciton. ? By that I mean, does the ARM approach with a RISC instruction set simply have a lower number of possible next instructions compared to and x86 CISC approach ? ( I genuinely dont know )

I.e.
If ARM has an average of 5 possible branch choices, and x86 has 6 branch choices ... by the law of averages, the ARM will better it'll have a 20% chance of being right (at complete random choice) versus 16% for x86. ?
 
Donnie Fisher said:
Could that be because RISC has less possible branch prediction routes that it can go down, so with less options to choose from, it automatically becomes better at prediciton. ? By that I mean, does the ARM approach with a RISC instruction set simply have a lower number of possible next instructions compared to and x86 CISC approach ? ( I genuinely dont know )

I.e.
If ARM has an average of 5 possible branch choices, and x86 has 6 branch choices ... by the law of averages, the ARM will better it'll have a 20% chance of being right (at complete random choice) versus 16% for x86. ?
Click to expand...

Branch choices are determined by the code that's running, ISA is not that relevant here. It's about microarchitecture.

Five years ago, Apple A9 used to have a 5x branch misprediction percentage compared to Intel CPUs of the time, and now it has about 25% less. That's only because one platform got better every year and Intel just stagnated for the last 5 years.
 
Last edited:
smilingcrow said:
The architecture is built around an ISA so I imagine there must be room for some differences?
Click to expand...

The CPUs don't directly run x86 (or Aarch64) instructions, they have their own internal ISA, and the microarchitecture is built around that. In a way, that's part of the microarchitecture too. We don't know much about those, so difficult to say anything either way. AMD for example used to be bad at branch prediction, but is now also better than Intel, on the same ISA.

What's important here is the trend, and how better Intel was compared to earlier Apple CPUs at branch predicion, and how that gap was closed down and now Apple leads. It wasn't ARM ISA that changed that much, it was Apple's microarchitecture (also AMD's) improving every year while Intel decided to do nothing.
 
Another Cinebench R23 result:

https://www.youtube.com/watch?v=yGY-l-fKUXE

Admittedly not the best test environment, looks like the MBP is being choked to death on that rug :D

Macbook Pro:
  • Single core: 1493
  • Multi core: 7783
Macbook Air:
  • Single core: 1493
  • Multi core: 7032
These run for 10 minutes, so a measure of sustained performance, and in case anyone wondererd, yes, they can sustain performance, even the fanless Macbook Air does very well here. In fact if that's all the thermal throttling that you get from the fanless version, it jus shows that there is a lot of headroom in these chips to go faster which Apple has decided not to utilise, perhaps for the sake of battery life.

And yes, 100% competitive with Intel and AMD.

Still on the macOS 10.0 though (10.0.1 overclocks the CPU by 100MHz apparently).

Armageus said:
Shocker as M1 using Hardware acceleration beats Xeon not using Hardware acceleration :confused:
Click to expand...

Final Cut Pro heavily uses GPU acceleration, which is also be available on the AMD Vega GPU. I think the bottleneck on the iMac was GPU memory, with only 8GB for the Vega 56. If the load exceeds that, the transfer of data between system and GPU memory adds a lot of latency.
 
Last edited:
Some users report Cinebench R20 (emulation through Rosetta) gives the scores of 450 and 2300 for single and multicore performance. Roughly as fast as the Core i7 in existing 13-inch Macbook Pros.

Apple will lift the embargo in 2021 apparently :D
 
Anandtech review of M1:

https://images.anandtech.com/graphs/graph16252/119145.png

https://images.anandtech.com/graphs/graph16252/119372.png
​
https://www.anandtech.com/show/16252/mac-mini-apple-m1-tested

Conclusion:

The performance of the new M1 in this “maximum performance” design with a small fan is outstandingly good. The M1 undisputedly outperforms the core performance of everything Intel has to offer, and battles it with AMD’s new Zen3, winning some, losing some. And in the mobile space in particular, there doesn’t seem to be an equivalent in either ST or MT performance – at least within the same power budgets.​
 
Last edited by a moderator:
SPEC2017 (int and fp) benchmarks from Anandtech:

Single core conclusions:
In the overall new SPEC2017 int and fp charts, the Apple Silicon M1 falls behind AMD’s Zen3 in the integer performance, however takes an undisputable lead in the floating-point suite.

Compared to the Intel contemporary designs, the Apple M1 is able to showcase a performance leap ahead of the best the company has to offer, with again a considerable strength in the FP score.

While AMD’s Zen3 still holds the leads in several workloads, we need to remind ourselves that this comes at a great cost in power consumption in the +49W range while the Apple M1 here is using 7-8W total device active power.​

Multi core conclusions:
In the overall multi-core scores, the Apple M1 is extremely impressive. On integer workloads, it still seems that AMD’s more recent Renoir-based designs beat the M1 in performance, but only in the integer workloads and at a notably higher TDP and power consumption.

Apple’s lead against Intel’s Tiger Lake SoC at 28W here is indisputable, and shows the reason as to why Apple chose to abandon their long-term silicon partner of 15 years. The M1 not only beats the best Intel has to offer in this market-segment, but does so at less power.

...

Overall, Apple doesn’t just deliver a viable silicon alternative to AMD and Intel, but actually something that’s well outperforms them both in absolute performance as well as power efficiency. Naturally, in higher power-level, higher-core count systems, the M1 can’t keep up to AMD and Intel designs, but that’s something Apple likely will want to address with subsequent designs in that category over the next 2 years.
​

Looks like AMD does just edge it out in terms of overall performance, at a notably higher power consumption. Intel is not competitive with either of them.

Single core:

https://images.anandtech.com/graphs/graph16252/119342.png

https://images.anandtech.com/graphs/graph16252/119343.png

https://images.anandtech.com/graphs/graph16252/117493.png


Multi core:


https://images.anandtech.com/graphs/graph16252/119353.png

https://images.anandtech.com/graphs/graph16252/119355.png

https://images.anandtech.com/graphs/graph16252/119365.png

** Do Not Hotlink Images **
 
Last edited by a moderator:
4K8KW10 said:
@HACO Have they got performance per watt comparisons where all tested CPUs are at equal watts, so we can see watt-watt?
Click to expand...

No, that's a very tough benchmark to do so they probably didn't have it ready for the time of embargo lift. They do mention average power draw for each benchmark though.

I do expect it to come though at some point.
 
The m1 looks promising. Yeah I'd love to know how much its using when it's rendering. Is it reported as 31W?

Interesting to see where Apple take this, I have no idea how they will scale this into workstation machines, but i'll be very intrigued.
 
Last edited:
I'm looking at this and I can't help but look for the flaw. There's no such thing as a free lunch, and I'm looking for the trade off. Anyone got any ideas where it might be?
 
q974739 said:
I'm looking at this and I can't help but look for the flaw. There's no such thing as a free lunch, and I'm looking for the trade off. Anyone got any ideas where it might be?
Click to expand...
No Bootcamp, no VMware Fusion (yet), no external GPU support, two fewer TB3 ports, RAM limited to 16GB on the current models to name a few.
 
You must log in or register to reply here.
Back
Top Bottom