I have been waiting for 2020 to come for 7 years. No, not the pandemic, but the part where Apple releases Apple Silicon. I remember telling a classmate that we will have the same mobile silicon as the then iPhone 5s in our computers in the future.
That prediction came true last year with the Microsoft Surface Pro X. However, it did not make the same kind of splash that Apple made with the Apple Silicon M1 chip found in the MacBook Air, Pro and Mac Mini.
Although Apple’s current lineup is in a transitional phase, rather than an overnight move away from Intel, it does look promising. Even more so with the inclusion of Apple’s Rosetta 2 technology.
Apple Silicon goes many steps further toward what I could have hoped for with a lot of accelerators such as the Apple Neural Engine. The benchmark numbers speak for themselves with the Apple M1 having a whopping 1743 single-core Geekbench score. For reference, AMDs best score comes to 1719 and Qualcomm’s Snapdragon 888 tops off at 1135 in single-core performance.
| Manufacturing Node | Multi-Core Score | Single-Core Score | |
| Apple M1 (4+4 core) | 5nm | 7708 | 1743 |
| AMD R9 5950X (16 core) | 7nm | 17287 | 1719 |
| QUALCOMM Snapdragon 888 (1+3+4 core) | 5nm | 3681 | 1135 |
What is interesting is that Apple delivers 44% of the multi-core performance of a 5950X at around 1/3 of the power. For reference, the M1 chip is half the core count of the 5950X and is lacking the benefit of SMT that Ryzen has.
| All core power | |
| Apple M1(4+4 core) | 31 watts |
| AMD R9 5950X (16 core) | 119 watts |
Data by Anandtech testing here and here.
Another interesting thing to note with ARM-based SOCs is that they are always on the cutting edge of manufacturing nodes such as TSMCs 5nm meaning that they also have the technological edge over traditional x86 processors are still on 7nm for AMD and 10nm for Intel.
On the graphics side of things, the M1’s GPU performance lands between the RX 560X and GTX 1650 according to Anandtech’s testing. But here is the interesting part: it does so with just 7 watts.
But Apple did one thing which many other OEMs can look at which is that they moved the memory onto the same package as the processor (System in a package). This, in turn, reduces the power needed by the processor to retrieve data.
This trick can be used by OEMs to reduce power draw and reduce the space on the board itself, which is at a premium in ultrabooks.
Perhaps an investment into custom silicon can be done by the likes of Samsung who manufactures the Exynos series of chips and Microsoft who has the financial capacity to include custom silicon as we saw with the XBox Series X.
It is to be noted that they are using LPDDR4X 4266MHz ram which is shared by the GPU and CPU. This is another optimization made by Apple’s silicon team, which pools both memories rather than having two separate pools, one dedicated to the GPU and the other to CPU. This is another simple yet clever method to reduce power.
This was only just the first generation of the ARM revolution from Apple. We are going to see many more generations with rumors of 5G and up to 128 core variants of Apple silicon being prepped within the next few years. Samsung is also working on an Exynos chip with Radeon Graphics technology which could make an appearance in the next Galaxy device. The future in the mobile space is going to get interesting.
