Apple M4 vs Apple M3 – How do Apple’s two silicons compare?

Apple introduced its latest processor, the Apple M4, with on startup on the new iPad Pro. This is the first time the company has packed a new chip into a tablet before on a MacBook. So, this goes without saying there is a lot of expectation about how much performance improvement it can bring when it is in a laptop. Let’s give you an idea of ​​it by comparing it to its predecessor, the Apple M3.

First, meet the Apple M4

The M4 marks the exciting fourth generation of Apple’s custom-designed M-series system-on-a-chip (SoC). It boasts a powerful processor with nine or ten coresdepending on the configuration. This processing muscle comes with powerful 10-core GPU to meet all your graphics needs. It also has a powerful 16-core Neural Engine. It is there to handle demanding tasks.

Currently, The 2024 iPad Pro is the only device rocking the Apple M4 chip. You can get it in two configurations. Built around a cutting-edge manufacturing process, the M4 offers a slight performance advantage over its M3 predecessor (we’ll get into that shortly).

The Apple M4 contains a whopping 28 billion transistors, a significant jump from the M3’s 25 billion. It also benefits from a wider memory bandwidth of 120 Gbps. So it gives you a 20 Gbps advantage over the M3.

Adding another layer of intrigue, Apple equipped the SoC with a dedicated display engine. Currently, this mechanism refines the visual effects of iPad Pro’s stunning Tandem OLED display. For all the graphics enthusiasts out there, The Apple M4 boasts hardware-accelerated ray tracing. It also has mesh shading capabilities.

Comparing the Apple M4 to its predecessor

This goes without saying The Apple M4 promises a significant leap over the M2 iPad Pro power supplies. But the real question remains: how does it fare with the M3, which has never graced the iPad lineup? While Apple hasn’t directly compared the two, let’s dive into their specs for a closer look.

Solid CPU performance improvement

Both The Apple M4 and M3 chips rely on a combination of cores for performance and efficiency. Performance cores provide lightning-fast speeds for demanding tasks, while efficiency cores superior in management lighter workloads while preserving battery life.

This is where the M4 starts to pull ahead. According to the benchmark results, the Apple M4 boasts a base clock speed of 4.40 GHz, surpassing the M3’s 4.05 GHz. Thiscombined with the potential for more cores in the M4, means a significant performance boost.

The Apple M4 comes in two variants. There is one with 9-core processor configuration, featuring 3 productivity cores and 4 performance cores. The second is a 10-core processor with an additional core for performance. On the other hand, the M3 only has features 8 cores. Half are dedicated for productivity, and the other half handles less intensive tasks.

Independent benchmarks paint a clear picture. Tom’s Guide compares the iPad Pro 2024 with M4 (10-core processor, 16 GB RAM) with the 13-inch MacBook Air with M3. The M4-powered iPad Pro dominated scoring huge 21% higher in single-core tests and a commendable 20% lead in multi-core tests.

Another benchmark discovered by MacRumors revealed that even the lower-end 9-core Apple M4 outperformed the M3 by roughly 13% in multi-core tasks. Although these results come from different devices, they provide valuable insight into the M4’s potential performance, which should lead to a smoother and faster user experience in real-world scenarios.

Productivity gains range from 13% to 20%, could, could looks modest on paper. However, these improvements should not be downplayed. The M4 provides a noticeable increase in power for demanding tasks such as video editing, 3D rendering and high-performance gaming.

There is also an improvement in GPU performance

While the Apple M4 boasts a significant processor jump over the M3, the graphics processing unit (GPU) story is a bit more nuanced. Apple claims the M4’s 10-core GPU is based on the architecture introduced with the M3. On paper, both GPUs appear quite a lot similar.

Both M4 and M3 offer 10 cores and support advanced features such as hardware-accelerated ray tracing and mesh shading, delivering improved renderings on the iPad for the first time (although these features were already available on the MacBook with the M3). M4 also introduces dynamic caching for optimized performance.

Apple advertises the M4’s GPU as creature four times faster rendering compared to the M2 chip. Based on this information, and considering the reported 1.8x speed increase of the M3 over the M2, a logical conclusion would be that the M4’s GPU delivers roughly double the performance of the M3.

However, real-world results can not perfect reflects these calculations. Early Metal benchmark results reported by 9to5Mac show a more modest 13% improvement for the M4’s GPU. This offers a gradual upgrade rather than a revolutionary leap.

On a brighter note, the M4’s media engine take a step forward with hardware-accelerated support for 8K video in H.264, HEVC, ProResand ProRes RAW formats. This is a clear improvement over the capabilities of the M3 chip.

About the Neural Engine in the Apple M4

With the growing prominence of generative AI features, Apple has placed significant emphasis on M4’s Neural Engine. This powerhouse boasts 16 cores and is capable of a staggering 38 trillion operations per second (TOPS). By comparison, the M3’s 16-core Neural Engine reaches 18 TOPS. Apple confidently declares the M4’s Neural Engine to be “faster than the neural processing unit of any artificial intelligence computer today.”

However, a report from Tom’s hardware throws up a curve highlighting that Apple uses different computing standards to measure the performance of the M3 and M4’s Neural Engine. The M3 rating is based on FP16 (half-precision floating point), while the M4 uses INT8 (8-bit integer). When adjusted to a common INT8 standard, Tom’s Hardware suggests that the M4’s Neural Engine offers a modest 5% advantage over the M3.

Another report from MacWorld shows that the M4 achieves roughly a 12% advantage in the Geekbench ML tests over the base M3. It is important to note that The tested M3 probably had 16GB of RAM, so an 8GB variant may show lower performance.

In conclusion, the M4’s Neural Engine undoubtedly outperforms the M3’s, but the margin isn’t as dramatic as Apple’s initial claims suggest. M4 supplies a clear increasing productivity for AI-powered tasks, but for most users, the difference can no be deal breaker.