Benchmarks

[nsayer]

This is all old news, I know, but I decided to do some Geekbench runs on all of the things I have available for comparison. That includes

• iPhone 13 Pro
• M1 Mac Mini
• M1 Max MBP

The biggest surprise to me was that the single core scores were not very far apart from each other. 1730 for the phone, 1726 for the mini and 1754 for the MBP.

My understanding is that the cores in all 3 chips are essentially the same, and so the single core scores lining up should not be a surprise.

The multicore scores diverge in the manner one would expect - 4744 for the phone, 7632 for the mini and 12431 for the MBP. The mini having 4x4 and the MBP having 8x2 seems enough to explain a 40% better score - 20% more total cores and 2 of them trade efficiency for performance. The phone having 2 performance and 4 efficiency cores is 40% lower than the mini probably with the nearly the same math - two fewer performance cores must be enough for 40% less oomph in multi-core scoring.

 

[Gnattu]

GeekBench is not a good benchmark but only a fast benchmark, and it cannot measure the machine's performance under thermal pressure because the test duration is simply too short.

3 chips are essentially the same

The "cores" are essentially the same, not the "chips". The whole chips are very different in may ways.

 

[nsayer]

The "cores" are essentially the same, not the "chips". The whole chips are very different in may ways.

The actual line above reads

My understanding is that the cores in all 3 chips are essentially the same,

 

[Jorbanead]

The actual line above reads

The "cores" are essentially the same, not the "chips". The whole chips are very different in may ways.

You both are wrong though.

The cores are not the same. M1 is based off of the A14 chip, whereas the iPhone 13 has a newer A15 chip.

 

[nsayer]

You both are wrong though.

The cores are not the same. M1 is based off of the A14 chip, whereas the iPhone 13 has a newer A15 chip.

So what's been changed in the cores between those two generations? They seem to have about the same single core geekbench score, so I posit that the answer is "not much."

 

[dugbug]

So what's been changed in the cores between those two generations? They seem to have about the same single core geekbench score, so I posit that the answer is "not much."

Well in the A14 vs A15 comparison (which is all we can do right now) we can see about 8% single core increase and 10.5% in multicore (someone check me).

Here is anandtech's deep dive of the A15.

The Apple A15 SoC Performance Review: Faster & More Efficient

www.anandtech.com

-d

 

[mr_roboto]

So what's been changed in the cores between those two generations? They seem to have about the same single core geekbench score, so I posit that the answer is "not much."

The numbers you found actually show that A15 probably is substantially better than A14/M1. Here's a pair of A14 and M1 GB5 scores to show what I mean:

iPhone 12 Pro Max - Geekbench

Benchmark results for an iPhone 12 Pro Max with an Apple A14 Bionic processor.

browser.geekbench.com

Mac mini (Late 2020) - Geekbench

Benchmark results for a Mac mini (Late 2020) with an Apple M1 processor.

browser.geekbench.com

Same A14/M1 "Firestorm" performance core, significantly different single-core scores. Apple caps maximum clocks at lower values in iPhones. An A15 iPhone essentially tying M1 Mac single core scores shows progress.

 

[Jorbanead]

So what's been changed in the cores between those two generations? They seem to have about the same single core geekbench score, so I posit that the answer is "not much."

You can't just compare Geekbench single-core speeds and infer they are the same. We know 100% that they are not the same. M1 uses Firestorm and Icestorm cores. A15 uses Avalanch and Blizzard cores.

The M1 performance cores are clocked at 3.2GHz while the A14 is clocked at 3.0GHZ. This is why you see better single-core numbers on the M1 compared to the A14. Macs have a higher thermal envelope than phones so they can push the M1 more. Now the A15 uses a new generation that improves performance, so the A15 can reach similar single-core speeds to the M1 while utilizing less power on an even lower clock speed inside a thermally contained iPhone. The A15 is just much more efficient.

This is why M1 single-core is similar to A15 single-core on Geekbench. A15 is on a new generation.

 

[gradi]

I watched this video today. Of course, it is speculation, but probably close to correct with regards to performance.

M2 vs M1 Pro vs M1 Performance

If you want to skip past the first 3 minutes that is rumors of what will be announced on March 8th I have it cued up to the beginning of the performance comparison stuff: