Apple unveils its new M4 chip

Apple has unveiled the M4, the primary chip in its fourth technology of Apple Silicon processors.

The M4, because of ship this month within the new iPad Pro, combines the hardware-accelerated ray tracing capabilities of Apple’s present M3 chips with enhanced AI capabilities.

Apple claims that the M4’s revamped Neural Engine is “extra highly effective than any Neural Processing Unit in any AI PC in the present day”.

Under, we’ve summarized its key specs, and the way they might have an effect on efficiency of CG software program.

Entry-level Apple Silicon chips by technology
M4 M3 M2 M1
Course of 3nm 3nm 5nm 5nm
Unified reminiscence (1,2) 16 GB
(24 GB)
24 GB 24 GB 16 GB
Reminiscence bandwidth 120 GB/s 100 GB/s 100 GB/s (68Gb/s)
CPU cores 10 8 8 8
GPU cores 10 10 10 8
GPU compute (2)
TBC TBC 3.6 (2.6)
{Hardware}-accelerated ray tracing? Sure Sure No No
Neural Engine cores 16 16 16 16
AI efficiency (2)
38 18 (16) (11)
Video decode engines 1 1 1 1
Video encode engines 1 1 1 1
ProRes encode/decode engines 1 1 1 0
AV1 decode Sure Sure No No
Launch date 2024 2023 2022 2020

(1) Most configuration obtainable
(2) Figures in brackets are from third-party web sites

Iterative enhancements to key specs, however an enormous increase to AI workloads
It’s solely just a little over 5 months since Apple introduced its milestone M3 family of chips, which introduced hardware-accelerated ray tracing to Macs for the primary time.

Compared, the M4 feels extra like an iterative replace, enhancing present capabilities quite than introducing new ones.

CPU core rely rises from 8 to 10 – in Apple’s terminology, each of the brand new cores are ‘effectivity’ quite than ‘efficiency’ cores – whereas GPU core rely and structure stays unchanged.

Reminiscence bandwidth additionally rises, from 100GB/s to 120GB/s, though the utmost RAM capability obtainable for an M4 chip within the new iPad Pros is simply 16 GB.

That’s really down on the M3, though that could be as a result of the M3 is obtainable within the a lot higher-powered MacBook Pro.

Like its predecessor, the M4 helps hardware-accelerated ray tracing, and its media engine helps AV1 decoding, in addition to H.264, HEVC and ProRes decoding and encoding.

The principle advance is in Neural Engine, the chip’s Neural Processing Unit: whereas core rely stays unchanged, efficiency jumps from 18 to 38 trillion operations per second.

There are caveats to a few of that, as Apple hasn’t launched quite a lot of supporting element on the brand new chip. In the event you’re , AnandTech has an in depth evaluation of the M4’s structure.

Apple’s efficiency comparisons between a 13-inch M4 iPad Professional and the closest equal M2 mannequin. CPU efficiency was measured utilizing Affinity Picture; 3D rendering with Octane X.

Benchmark comparisions for CG software program
So how do these specs translate into real-world efficiency in CG software program?

Apple’s foremost particular declare is that rendering in Octane X 2024.1 is “as much as 4x” sooner on a 13-inch iPad Professional with a 10-core-CPU M4 than the equal previous-generation mannequin, which has a 8-core-CPU M2 chip.

For CPU efficiency, the built-in benchmark in Affinity Picture 2.4 is “as much as 1.5x sooner”.

Surprisingly, Apple hasn’t quantified the influence of the M4’s Neural Engine on AI workloads.

Though an M4 iPad Professional can “isolate a topic from its background all through a 4K video in Ultimate Reduce Professional with only a faucet”, the agency doesn’t say how a lot sooner it does so than a previous-gen M2 iPad Professional.

Worth and launch date
M4 chips are because of ship within the new iPad Execs, at the moment obtainable to pre-order, and due in shops on 15 Could 2024.

Pricing for the brand new iPad Professional begins at $999 for an 11-inch WiFi-only mannequin with 256GB storage, and $1,299 for a WiFi-only 13-inch mannequin, once more with 256GB storage.

Read more about the M4 chip on Apple’s website

Have your say on this story by following CG Channel on Facebook, Instagram and X (formerly Twitter). In addition to having the ability to touch upon tales, followers of our social media accounts can see movies we don’t put up on the location itself, together with making-ofs for the most recent VFX films, animations, video games cinematics and movement graphics tasks.