Home
Computer Hardware + Systems Articles!

Apple M1 : How Fast Are Its CPU + GPU?

Posted by
Date:
Spread the love

The Apple M1 is the first ARM-based SoC (System on a Chip) designed by Apple to be used as the CPU for Mac computers.

Here is EVERYTHING you need to know about the Apple M1, including how fast its CPU and GPU really are!

 

Apple M1 : First ARM SoC For Mac Computers!

The Apple M1 is the first ARM-based SoC (System on a Chip) designed by Apple for use in Mac computers, debuting in :

  • 2020 MacBook Air (4th Gen) : US | UK | AU | MY | SG
  • 2020 MacBook Pro (6th Gen) : US | UK | AU | MY | SG
  • 2020 Mac mini (5th Gen) : US | UK | AU | MY | SG

Packing 16 billion transistors, it is the first chip to be manufactured on the new 5 nm TSMC process technology.

As an SoC, it combines 8 CPU cores, an 8-core GPU, a 16-core Neural Engine, an image signal processor (ISP), an NVMe storage controller, a Thunderbolt 4 controller and a Secure Enclave – all in a single chip.

The Apple M1 is paired with either 8 GB or 16 GB of LPDDR4X memory using a system-in-a-package design.

Apple M1 : Everything On First ARM CPU For Mac Computers!

8-Core CPU : ARM big.LITTLE Design

The Apple M1 uses the ARM big.LITTLE design for the CPU. Its 8-core CPU is divided into two performance clusters :

  • P Cluster (Performance) : 4 x Firestorm cores (up to 3.2 GHz)
  • E Cluster (Efficient) : 4 x Icestorm cores (up to 2.06 GHz)

The P Cluster is designed for high-performance, with a power consumption of 13.8 watts. It has a large 12 MB shared L2 cache, with a 128 KB dedicated L1 cache for each Firestorm core.

The E Cluster, on the other hand, is designed for a low power consumption of just 1.3 watts. It has a smaller 4 MB shared L2 cache, with a 64 KB dedicated L1 cache for each Icestorm core.

8-Core GPU : Mysterious Indeed!

The M1 uses an Apple-designed 8-core GPU, of which very little is known. Apple only says that it can handle up to 25,000 concurrent threads.

But from what we can tell, it has 128 execution units with an FP32 performance of 2.6 teraflops per second, a 82 GT/s texture fill rate, and a 41 GP/s pixel fill rate.

Apple M1 cores

16-Core Neural Engine

The Apple M1 features a 16-core Neural Engine, which is capable of executing 11 trillion operations per second.

LPDDR4X Unified Memory

The Apple M1 uses a unified memory design, which means the CPU and GPU share the same memory.

Apple offers 8 GB and 16 GB LPDDR4X SDRAM options, built directly onto the package for near-instantaneous access.

Rosetta 2 Translation

As most Mac apps are compiled for x86 processors, they need to be translated to work with the Apple M1, which uses the totally different ARM architecture.

That’s where Apple Rosetta 2 comes in – introduced in macOS Big Sur, it translates or emulates x86 apps so that they can run properly on the M1 processor.

Certain apps, especially games, may require third-party software like Wine or CrossOver to work properly, like this example of The Witcher 3 running on the Apple M1.

 

Apple M1 : Specifications

Apple is hardly the most transparent company in the world, but here is what we know so far about the M1 chip’s specifications :

Specifications Apple M1 (APL1102)
Fab Process 5 nm TSMC
Transistor Count 16 Billion
Die Size 120 mm²
CPU Cores P Cluster : 4 x Firestorm cores
E Cluster : 4 x Icestorm cores
L1 Caches Firestorm : 128 KB per core
Icestorm : 64 KB per core
L2 Caches P Cluster : 12 MB shared
E Cluster : 4 MB shared
Clock Speed P Cluster : 0.6~3.204 GHz
E Cluster : 0.6~2.064 GHz
CPU Power
Consumption
 P Cluster : 13.8 W
E Cluster : 1.3 W
GPU Cores 8 Cores, 128 Execution Units
GPU FP32 Performance 2.6 TFLOPS
GPU Texture Fillrate 82 GT/s
GPU Pixel Fillrate 41 GP/s
Memory Size 8 GB / 16 GB
Memory Type LPDDR4x-4266
Memory Interface 128-bit
Memory Bandwidth 68.27 GB/s
Neural Engine 16 Cores
Neural Engine
Performance
 11 Trillion Ops / Second
TDP 10 watts (MacBook Air)

 

Apple M1 : How Fast Are Its CPU + GPU?

Unfortunately, we do not have access to the Apple M1, but Ars Technica tested it in the 2020 Mac mini. Take a look at some of the results they posted :

Note : While Ars Technica listed the M1 as a 10 watt part, this is only what Apple claims for the MacBook Air. It is likely to be clocked higher, with a higher TDP in the Mac mini and MacBook Pro.

Geekbench 5 : Single-Threaded CPU Test

Will you look at that? The Apple M1’s Firestorm cluster has really good single-core performance!

It was 7% faster than the Core i7-1185G7, in both 28W and 15W TDP modes; and 53% faster than the Ryzen 7 4700U!

Geekbench 5 : Multi-Threaded CPU Test

This is even more astounding! The Apple M1 is even faster in the multi-threaded test!

In the multi-threaded test, the Apple M1 was 22% faster and 54% faster than the Core i7-1185G7 in 28W and 15W modes, respectively. It was also 52% faster than the Ryzen 7 4700U!

Geekbench 5 : GPU Test

Even its 8-core GPU did very well, beating the Core i7-1185G7‘s integrated Iris Xe by 26%, and Ryzen 7 4700U‘s integrated Vega 7 by 65%.

Cinebench R23 : Single-Threaded Test

The Apple M1 was as fast as the 28-watt Core i7-1185G7 in the single-threaded test, and was 28% faster than the Ryzen 7 4700U. And it was only 6.5% slower than the Ryzen 9 5950X.

Cinebench R23 : Multi-Threaded Test

In the multi-threaded test, the Ars Technica team set a limit of 8 threads for a level-playing field.

The Apple M1 did well, beating the Ryzen 7 4700U by 14%, and the Core i7-1185G7 by 25%. And it was just 34% slower than the Ryzen 9 5950X.

 

Apple M1 Mac Computers : Where To Buy?

The Apple M1 is currently available in these Mac computer models :

  • 2020 MacBook Air (4th Gen) : US | UK | AU | MY | SG
  • 2020 MacBook Pro (6th Gen) : US | UK | AU | MY | SG
  • 2020 Mac mini (5th Gen) : US | UK | AU | MY | SG

 

Recommended Reading

Go Back To > Gaming | Computer | Home

 

Support Tech ARP!

If you like our work, you can help support us by visiting our sponsors, participating in the Tech ARP Forums, or even donating to our fund. Any help you can render is greatly appreciated!


Tags:

About The Author

Leave a Reply