Apple has announced the M4 chip, its successor to the M3 that’s currently found in a bunch of Macs and iPads. The M3 is an excellent chip and a real leap above the M2 that came before it, so the question is whether the M4 can manage a similar feat.
Right now, the M4 is only in the iPad Pro, and that means information about how good it is and what it does is rather limited. But if you’re interested in finding out more, you’re in luck, as we’ve gathered up everything we know about Apple’s M4 chip and compared it side by side with the M3. If you want to learn more about Apple’s next chip — and how it compares to the M3 — read on.
Where can you find these chips?
At the time of this writing, the M4 chip is only present in the 2024 iPad Pro. That device has a few different configurations of the M4, including with varying amounts of memory and storage, but right now it’s not found in any other Apple devices.
That said, the M4 is undoubtedly also coming to the Mac. Given it’s an entry-level chip, we can expect to see it in the 14-inch MacBook Pro, both sizes of the MacBook Air, the iMac, and the Mac mini. It’s expected to land in the MacBook Pro, the iMac, and the Mac mini around late 2024, with the MacBook Air following in spring 2025.
As for the M3 chip, that’s already out in the wild. It’s currently in the 14-inch MacBook Pro, the 13-inch and 15-inch MacBook Air, and the 24-inch iMac.
Unlike the M4, the M3 has never appeared in any iPad. The new iPad Air and previous-generation iPad Pro have the M2, while the iPad and iPad mini use the A14 and A15 chips, respectively.
What’s different under the hood?
There are a few differences under the hood between the M3 and M4 chips, but those differences are subtle rather than monumental. They’re less significant than the changes experienced when Apple moved from the M2 to the M3, but we won’t know for sure what kind of divergence there is until we get to test an M4-equipped device.
Both the M3 and the M4 are made using a 3nm manufacturing process, which is one reason for the more minor changes (the M2 was a 5nm chip, and moving to 3nm enabled Apple to radically improve performance and efficiency). However, it’s worth noting that the M4 is a second-generation 3nm chip, which means there are some small improvements to its efficiency and output.
One area where we can see an immediate difference is the number of cores each chip comes with. The M3 has an 8-core CPU made up of four performance cores and four efficiency cores. This is supplemented by either an 8-core or a 10-core GPU (depending on the model) and a 16-core Neural Engine.
The M4, on the other hand, has either a 9-core or 10-core CPU (with either three or four performance cores and six efficiency cores), a 10-core GPU and a 16-core Neural Engine. That means it’s a step up of only a couple cores here and there, but any increase in performance is definitely welcome.
In terms of the Neural Engine, Apple says this can perform 38 trillion operations per second (TOPS), making it “more powerful than any neural processing unit in any AI PC today.” The M2, for comparison, could reach 15.8 TOPS, while the M3 reportedly hit 18 TOPS.
That looks like a huge win for the M4, but it’s not quite the whole story, as per Tom’s Hardware. According to that outlet, the M4’s score is 38 TOPS with the INT8 data type. The M3, on the other hand, was 18 TOPS at FP16 precision. Once those two scores are equalized to INT8, the difference is about 5% in favor of the M4 — far less than the apparent doubling between the two chips.
There are a few other contrasts of note. The M4 is packed with 28 billion transistors compared to the M3’s 25 billion, and it has a memory bandwidth of 120Gbps compared to its predecessor’s 100Gbps. There’s also a dedicated display engine in the M4 that was absent in all of Apple chips before it, and this engine is used to handle the OLED display in the new iPad Pro.
Performance
It’s hard to assess performance differences between the M4 and the M3 right now, as we haven’t had a chance to review any device outfitted with the M4. Apple’s marketing materials also do not put the M4 up against the M3, presumably because the company is still selling M3 devices and doesn’t want to undercut them. The iPad Pro’s M4 is instead compared to the M2. When the M3 launched, meanwhile, Apple squared it up against the M1, so we can’t even see how the M4 and M3 compete against a common chip in the form of the M2.
We have had a small clue about the M4’s performance, however. Recently, the M4 chip turned up on benchmarking website Geekbench. There, the M4 achieved a score of 9,234 in the website’s ML 0.6.0 test, which measures machine learning capabilities. That’s about 23% higher than the M2 and around 10% better than the M3. So we’re looking at an incremental change rather than a seismic one. It’s worth noting, though, that Geekbench’s tests are synthetic benchmarks rather than real-world ones, so they may not translate exactly into actual usage.
It’s also hard to draw conclusions about Mac performance based on iPad test results (both synthetic and real world), as they are very different devices. For example, the MacBook Pro has active cooling, so it will likely perform better than the passively cooled iPad Pro and MacBook Air. MacBooks also have more space inside their shells, which helps with cooling (and, therefore, performance). For those reasons, it’s best to wait for reviews to draw any conclusions about the performance of the M3 or M4 device you’re thinking of buying.
One thing that will really help users of the M4 chip is its inclusion of Dynamic Caching, mesh shading, and hardware-accelerated ray tracing. These are not new to Apple users — they’re already in the M3 series — but the iPad Pro didn’t have these features when it was outfitted with the M2. So, while that’s not an advantage the M4 has over the M3, it’s worth considering if you’re looking at getting an M4 iPad Pro.
