Production Expert

View Original

Apple Silicon: A Guide To Choosing The Right Mac For Your Studio

Choosing the correct Mac computer for your studio can be very difficult. Which model should I choose, how many processors, how much memory, and so much more? In this guide, we will help you navigate all the options to help you choose the best Mac for your studio.

Which Apple Silicon Chip?

With the release of Apple Silicon, Apple switched from Intel processors with separate RAM and graphic processors to their own Apple silicon chips, which brings all these elements into one chip. This development produces significant performance benefits but at the expense of being unable to upgrade these elements.

To start with, we are going to look at the different Apple Silicon chips to determine which are best for music and audio applications and depending on which DAW you use, the answer may be different.

One of the challenges when choosing which computer is best for pro audio applications is that the normal benchmarks don’t tell the full story. To help, YouTuber James Zhan has undertaken some carefully controlled tests, comparing the performance of different DAWs running on an M1 Pro, M2 Pro and M3 Pro. These tests also prove helpful even if you are looking at Apple’s other chips like the Max and Ultra models.

The summary from James’ tests is that some DAWs can use the efficiency cores in Apple Silicon chips, and some cannot. This matters because as Apple has upgraded from 1st to 2nd and, more recently, to 3rd generation Apple Silicon chips, the number and ratio of performance to efficiency cores has changed with some models.

We have consolidated the key specifications of all the Apple Silicon chips into this one table to help you choose the most appropriate chip for your studio computer.

See this content in the original post

Notes:

  • Where there are multiple numbers in a cell, this shows the different options available for that spec.

  • * The specs for the M3 Ultra are predictions based on the fact that, so far, the Ultra is basically two Max chips put together.

  • Note that Unified Memory is what we used to refer to as RAM, but now as the RAM is built into the SoC and is shared with the GPU cores, it is referred to as Unified Memory.

The Number Of CPU Cores

As you can see from the image above, taken from James’ video, the different DAWs tested fall into two basic groups: those that can use the efficiency cores in Apple Silicon chips and those that cannot.

  • Those that can include Reaper, Cubase, and FL Studio.

  • Those that cannot include Pro Tools Logic Pro, Live, and Studio One.

In terms of CPU usage, if you plan to use Avid Pro Tools, Apple Logic Pro, Ableton Live or PreSonus Studio One, then you want to steer away from Apple Silicon chips with more efficiency cores and fewer performance cores, which means it won’t necessarily be the most recent Apple Silicon chip.

You can see from James’ tests that for DAWs like Avid Pro Tools, the best option within the Mx Pro range is the M1 Pro because it has the highest number of performance cores. Even though the M2 Pro and M3 Pro models have more cores overall, Apple has chosen to add more efficiency cores, which some DAWs, including Pro Tools, cannot currently use.

Using the results from James’ video, looking at the Mx Max range, it could be worth considering the M3 Max over the M1 Max or M2 Max because, unlike the Mx Pro range, the M3 Max has more performance cores over the M1 Max or M2 Max. But note that both the M2 Max and M3 Max also have double the efficiency cores, which some DAWs currently cannot harness.

Moving onto the Mx Ultra range, things are not as clear because until Apple releases a computer with an M3 Ultra chip, we cannot be sure what the M3 Ultra specs will look like. What we do know is that the M2 Ultra has the same number of performance chips as the older M1 Ultra chip. This means that from a CPU perspective, there is no benefit in getting a Mac computer with an M2 Ultra chip in it, but there still may be reasons to go for the M2 Ultra, for example, if you want more unified memory.

Finally, James’ research shows that there is a real benefit in choosing a Mac with an Apple Silicon chip with as many performance cores as you can afford.

The Number Of GPU Cores

The reality of the graphics processing power that Apple has built into their Apple Silicon chips is that for audio applications, there is way more GPU power than we are ever going to need. For example, my Intel Mac mini has an Intel UHD Graphics 630 graphics processor that shares memory with the system. I am driving two 3840x2160 screens without any issues. When it comes to Apple Silicon, we are seeing audio post users running without issues with a 30-core GPU. There is no need to go to the top-spec chips just for GPU performance.

One factor we do need to consider is how many screens an Apple silicon SoC can support. For example…

  • The MacBook Pro with an M3 chip can drive one external high-resolution display.

  • The MacBook Pro with an M3 Pro chip can drive up to 2 external high-resolution displays.

  • The MacBook Pro with an M3 Max chip can drive up to 4 external high-resolution displays.

As you can see, your choice of Apple silicon chip could be determined by the number of screens you need in your studio.

The Number Of Neural Engine Cores

It’s early days in audio and music production software when it comes to what software uses the neural engine.

  • We understand that Final Cut Pro’s Smart Conform functionality, which employs machine learning to reframe clips automatically, is noticeably faster on Apple Silicon Macs.

  • The Neural Engine in DaVinci’s Resolve run three times faster on Apple Silicon powered computers as it can harness the internal neural engine in dozens of tools that rely on accurate automated analysis of motion and content, such as Smart Reframe, SpeedWarp or SuperScale. It can classify audio clips based on their content, making editing choices faster when reviewing unfamiliar materials.

  • We know that audio restoration software Hush and Hush Pro can use the Neural Engine on Apple Silicon chips to accelerate rendering, enable real-time previews, and keep CPU usage to a minimum. Although Hush works on Intel machines, too, rendering may take significantly longer.

Our advice is not to choose an Mx Ultra chip just because it has a larger neural engine, but rest assured that more and more audio software is going to use Apple’s Neural Engine as the use of machine learning expands.

That said, there is no doubt that Machine Learning and Artificial Intelligence tools are a growth area, and developers are going to develop them to use the Neural Engine in Apple’s SoCs.

How Much Unified Memory To Go For?

For this, we turn to users’ experiences to guide us to the appropriate sweet spot. For example…

  • A user reports with an Apple Silicon computer with 32GB of unified memory, undertaking large file operations in Pro Tools takes ‘forever’, but when doing the same task with a computer with 64GB of unified memory, it does it in just ‘a minute’.

  • Meanwhile, a well-known audio post professional has an Apple M2 Max with a 12‑core CPU, 30‑core GPU, and 16‑core neural engine with 64GB of unified memory. With this computer, he can run a 500+ track Dolby Atmos mix, running the Dolby Atmos Renderer on the same machine, with the memory usage hovering around 35%.

  • User tests with Pro Tools appear to show that unlike on Intel-based machines, where it was beneficial to load as much of the session into RAM as possible, because of the SoC architecture and the increased memory bandwidth, setting the RAM buffer to Auto and allowing the software to move segments of the session into memory on the fly works fine.

Our advice is to go for at least 64GB of unified memory for most professional audio applications. However, if you are using a lot of virtual instruments which load their samples into memory then moving up to 96GB would make sense.

Does Memory Bandwidth Matter?

To answer this question, let’s first look at what memory bandwidth is. Wikipedia defines memory bandwidth as…

“Memory bandwidth is the rate at which data can be read from or stored into a semiconductor memory by a processor. Memory bandwidth is usually expressed in units of bytes/second, though this can vary for systems with natural data sizes that are not a multiple of the commonly used 8-bit bytes.”

When Apple announced the M2 chip during the WWDC 2022 keynote, it made a big thing of increasing the memory bandwidth from the M1 to M2, with the M1, which increased memory bandwidth to 100GB/s. The Mx Pro range offers 200GB/s, the Mx Max range 400GB/s and the Mx Ultra range 800GB/s.

But do we need all this memory bandwidth? Our research turned up an article Apple's M1 Pro, M1 Max SoCs Investigated: New Performance and Efficiency Heights by Andrei Frumusanu, written back in 2021. Andrei sets about trying to see what uses all this memory bandwidth. After all, surely Apple wouldn’t have spent the money creating all this memory bandwidth if it didn’t need to? Firstly, investigating how much memory bandwidth the M1 Max CPU uses gets to a figure of 243GB/s, but this is only just over half the available memory bandwidth for the Mx Max range of chips. Andrei continues…

“While 243GB/s is massive and overshadows any other design in the industry, it’s still quite far from the 409GB/s the chip is capable of. More importantly for the M1 Max, it’s only slightly higher than the 204GB/s limit of the M1 Pro, so from a CPU-only workload perspective, it doesn’t appear to make sense to get the Max if one is focused just on CPU bandwidth.

That begs the question, why does the M1 Max have such massive bandwidth? The GPU naturally comes to mind. However, in my testing, I’ve had extreme trouble finding workloads that would stress the GPU sufficiently to take advantage of the available bandwidth. Granted, this is also an issue of lacking workloads, but for actual 3D rendering and benchmarks, I haven’t seen the GPU use more than 90GB/s (measured via system performance counters). While I’m sure there’s some productivity workload out there where the GPU is able to stretch its legs, we haven’t been able to identify them yet.

That leaves everything else, which is on the SoC, media engine, NPU, and workloads that would simply stress all parts of the chip at the same time. The new media engine on the M1 Pro and Max are now able to decode and encode ProRes RAW formats. With a 5K 12bit sample with a bitrate of 1.59Gbps, the M1 Max is not only able to play it back in real-time, it’s able to do it at multiple times the [real time] speed, with seamless immediate seeking. Doing the same thing on my 5900X [Intel] machine results in single-digit frames. The SoC DRAM bandwidth while seeking around was at around 40-50GB/s – I imagine that workloads that stress CPU, GPU, and media engines all at the same time would be able to take advantage of the full system memory bandwidth and allow the M1 Max to stretch its legs and differentiate itself more from the M1 Pro and other systems.”

In summary, Andrei found…

  • The CPU cores can only use 243GB/s of memory bandwidth

  • They could not find any scenario to load 400GB/s bandwidth fully with CPUs and GPUs alone.

  • It was only together with media engines and other parts of the chip that it was possible to get close to utilizing the 400GB/sec memory bandwidth.

To give you a reference point, the memory bandwidth in the M1 base model already offers twice the memory bandwidth of an Intel chip powered computer.

It is suggested that as music production software is driven mostly by CPU usage, you are unlikely to benefit from 400GB/sec bandwidth unless you are using very high-quality video files, which could stretch the media engine.

In conclusion, don’t let the memory bandwidth specs influence your choice of Apple computer chip for audio applications.

Which Chip Should I Go For?

This is where you will need to use all the information in this article to determine which Apple Silicon model to choose. It will be different depending on which DAWs you use.

If you use Pro Tools, Logic Pro, Live, and Studio One, which currently don’t use the efficiency cores, choose a chip with as few efficiency cores as possible.

If they include Reaper, Cubase, and FL Studio, which make use of the efficiency cores, then the choices are more flexible, although we would suggest getting as many performance cores as you can afford.

The only specific advice we would offer is not to go for a Mac computer with the base model SoC. The main reason is the lack of unified memory, with the maximum possible amount being only 24GB. In addition, the number of cores is limited, and the limited number of screens you can have may be an issue for some.

Which Model? Desktop Or Laptop

Now, we move on to look at everything that surrounds the System on a Chip. Here, your choices will be largely determined by how you will be using your computer and what computer-related I/O you will need.

The first area to consider is whether you want a desktop or laptop computer. If you need portability, then a laptop is the obvious route, but be aware you are effectively paying a premium for choosing a laptop. But if your laptop could double up as your desktop when you are in your studio by using some form of docking station, then that could be a real win-win.

If you don’t need portability or you need a computer permanently in your studio, then a desktop computer is the obvious choice.

Which desktop? Mac mini, Mac Studio, Mac Pro

In our opinion and experience, the Mac Studio, whether you go for the Max or Ultra chips, is the sweet spot. With the Mx Max, you get four Thunderbolt 4 USB-C ports for your peripherals. With the Mx Ultra, you get four Thunderbolt 4 USB-C ports on the back and two more on the front for your peripherals. As to which generation of Apple silicon chip you go for refer to the CPU section above to help you choose.

If you want a lower-cost option, then a Mac mini with an Mx Pro chip could be an option. Note that once you go for the Mx Pro option, you will get three Thunderbolt 4 USB-C ports for your peripherals.

When it comes to the Mac Pro, there are a limited number of reasons that would make it worth going to a Mac Pro, all of which we outline in detail in our article Which to Buy - Apple Mac Studio or Mac Pro 2023?

Which Laptop? MacBook Air, MacBook Pro

Your choices when it comes to a laptop, taking into consideration the limits of the base model SoCs, leaves you looking at an Mx Pro or Mx Max 14-inch or 16-inch models, all of which come with three Thunderbolt 4 USB-C ports for your peripherals. As to which generation of Apple silicon chip you go for refer to the CPU section above to help you choose.

Can I Still Get Older Models?

Obviously, you can pick up a secondhand model with older Apple Silicon chips but do check out the Apple Refurbished section of the Apple Store, as Apple are still selling refurbished M1 and M2-powered Macs. A refurbished M1 or M2 Mac can offer significant savings without substantially compromising the performance quality necessary for professional audio production. When considering the performance and price difference between these models and the latest M3 Macs, the savings could be better allocated towards other things or just not spent at all!

Apple Refurb With AppleCare

The refurbished world is not what it used to be. Apple's certified refurbished products undergo rigorous testing to ensure they meet high standards. Purchasing a refurbished M1 or M2 Mac means you receive a device in 'like new' condition but at a more attractive price point. For those of us in the professional audio world industry, where budgets are decreasing for the same work, this is an opportunity to economise intelligently without skimping on quality.

What’s also worth noting is that an Apple Refurbished product qualifies for the Apple Care programme, which is well worth buying for peace of mind.

In Conclusion

There you have it. Depending on which DAWs you use, there may be little or no benefit in going for the latest Apple Silcon-powered models. You may be able to make significant savings by buying an earlier model with an M1 or M2 SoC.

See this content in the original post