Apple M5: Specs, RAM for VMs, and Windows Performance

The M5 generation gives Mac users more memory bandwidth, while M5 Pro and M5 Max provide additional resources for heavier workloads. Those improvements are important when macOS and a virtual machine are competing for the same hardware, but they don't mean you should automatically give Windows more RAM.

For most people running Windows 11 in Parallels Desktop, the best starting point is to leave CPU and memory allocation set to Automatic (Recommended). If you configure memory manually, the goal isn't to give Windows as much as possible. Your Mac still needs enough resources to run macOS and everything else you have open.

The difference between M4 and M5 also depends on what you do inside the VM. Office, email, and browser-based work aren't the same challenge as compiling a large project or running several VMs at once.

This guide looks specifically at those virtualization questions: what's changed from M4, how to think about VM memory, and what to expect when running Windows and Linux on an M5 Mac.

For a broader look at the chip itself, see our Apple M5 chip overview.

M5 vs. M4: What changed for virtualization

For virtualization, the key M5 upgrades over M4 are increased memory bandwidth and greater CPU capacity on M5 Pro and M5 Max. How much those changes improve VM performance depends on the workload.

More memory bandwidth

The base M5 reaches 153 GB/s of unified memory bandwidth, up from 120 GB/s on the base M4. M5 Pro reaches up to 307 GB/s, while M5 Max reaches as high as 614 GB/s depending on its GPU configuration.

Memory bandwidth matters because the host and guest operating systems share the same underlying hardware. Workloads that move large amounts of data through memory can put that bandwidth to use, particularly when you're doing substantial work in both macOS and a VM at the same time.

That doesn't mean you'll see a 30% improvement in Windows because memory bandwidth increased by roughly that amount on the base chip. VM performance depends on the application, resource allocation, storage activity, graphics requirements, and whether the Windows software is Arm-native or being emulated.

More CPU capacity at the Pro and Max tiers

The base M5 remains a 10-core CPU. The larger change comes higher in the lineup, with M5 Pro available with up to an 18-core CPU and M5 Max offering an 18-core CPU.

Extra cores can be useful for CPU-heavy guest workloads and for people who keep demanding Mac applications running while a VM is busy in the background. Developers compiling code inside Windows or Linux are more likely to notice that capacity than someone using a Windows VM for Outlook and Excel.

There is still a balancing act. Assigning every available CPU core to a VM doesn't make it faster. macOS needs CPU resources too, and taking too much away from the host can hurt performance on both sides.

Familiar memory ceilings

For the chips covered here, maximum unified memory remains 32 GB on M5, 64 GB on M5 Pro, and 128 GB on M5 Max.

That's important when choosing a Mac for virtualization, but total memory is only part of the decision. A 64 GB Mac doesn't mean a Windows VM should receive anything close to 64 GB by default.

The useful question is how much memory your actual guest workload needs.

How much RAM should you give a Windows VM on M5?

Start with Automatic (Recommended).

Parallels Desktop can automatically configure CPU and memory resources for your virtual machine. For most Windows users, there's little reason to change those settings before you've identified a workload that needs something different.

Giving Windows more memory isn't automatically better.

On an M5 Mac, macOS and your virtual machines draw from the Mac's unified memory. If you allocate too much to Windows, you can leave macOS short on resources. Since Parallels Desktop itself runs on macOS, a struggling host can affect the VM as well.

Allocate too little and you create the opposite problem: Windows and memory-heavy Windows applications may become less responsive.

If you do need to configure memory manually, Parallels recommends assigning no more than half of your Mac's memory to a virtual machine. When several VMs are running simultaneously, that guidance applies to their combined allocation.

Think about the workload before changing the number.

A Windows VM used for Office, a browser, or a business application has very different requirements from a development VM running an IDE, local services, databases, and build tools. Parallels Desktop 26 recommends 16GB or more of memory on the Mac itself for high loads, software development, graphics-intensive applications, and running multiple virtual machines.

The practical approach is straightforward: start with Automatic, use the VM normally, and change the allocation only when the workload calls for it.

What about the maximum VM memory?

The maximum memory you can assign also depends on your Parallels Desktop edition.

Parallels Desktop supports up to 8 GB of memory assigned to a VM in the standard edition, while Parallels Desktop Pro Edition supports VM memory allocations up to 128 GB. Your Mac's available memory and hardware still determine what can actually be assigned.

For most users, the maximum isn't a target. A VM should get the resources its workload needs while leaving enough capacity for macOS.

Windows performance on M5

Windows 11 on Arm is the version of Windows you'll normally run in Parallels Desktop on an M5 Mac. Because both Apple silicon and Windows 11 on Arm use the Arm architecture, Windows itself can run as an Arm guest rather than requiring x86-to-Arm emulation for the entire operating system.

Applications are a little more complicated. Arm-native Windows applications are the most direct fit. Many common Windows applications now have native Arm versions, and those don't need x86 or x64 instruction translation. Older x86 and x64 Windows applications can often run too. Windows 11 uses Microsoft's Prism emulation technology to translate their instructions for Arm. That translation happens inside Windows rather than in Parallels Desktop.

Compatibility can still vary. Microsoft notes that Windows on Arm emulation applies to user-mode code, not kernel-mode drivers. Software that depends on an incompatible x86 or x64 driver may therefore behave differently from an ordinary desktop application.

So what does M5 change?

For everyday Windows work, the difference may be less dramatic than the specifications suggest. Office applications, browsers, communication tools, and many business applications aren't likely to keep a high-end M5 Max busy.

Heavier workloads have more room to benefit. Compiling large projects, running development tools alongside demanding Mac applications, or keeping multiple VMs active can make greater use of the CPU and memory resources available on M5 Pro and M5 Max.

That's also why choosing between M5, M5 Pro, and M5 Max should start with the workload rather than the biggest number on the spec sheet.

Linux VM performance on M5

Linux is a particularly good fit for Apple silicon because major distributions already offer Arm versions.

Parallels Desktop supports Arm versions of Ubuntu, Fedora, Debian, Red Hat Enterprise Linux, Kali Linux, and other guest operating systems on Apple silicon Macs.

Using an Arm build means the guest operating system doesn't need x86-to-Arm instruction emulation. That's useful for developers who want a Linux environment for coding, testing, command-line tools, or debugging without maintaining a separate Linux machine.

The same resource rules still apply. A Linux VM has to share CPU and unified memory with macOS, so assigning more resources than the workload needs can be counterproductive. This becomes especially relevant when running several VMs. Look at their combined resource use, not each VM in isolation.

Is M5 better than M4 for virtual machines?

Yes. M5 offers more memory bandwidth than M4, while M5 Pro and M5 Max add more CPU capacity for demanding VM workloads.

How much of that you'll notice inside a VM depends on what the VM is doing. Someone who spends most of the day in Outlook, Excel, a browser, and a few Windows business applications may already have plenty of performance on an M4 Mac. Upgrading solely for those workloads is unlikely to reveal everything the newer hardware can do.

Development work is different. Large builds, several active VMs, memory-heavy tools, or demanding applications running simultaneously in macOS and the guest OS give the newer hardware more work to absorb.

It's also worth separating Apple's general M5 benchmarks from VM performance. Improvements measured in native macOS applications don't translate directly into the same percentage improvement inside Windows or Linux. Guest workload, VM configuration, application architecture, and emulation can all change the picture.

If virtualization is a major part of your day, the Pro and Max tiers become interesting less because they're “faster Macs” and more because they give demanding host and guest workloads more resources to share.

How Parallels Desktop runs Windows on an M5 Mac

On an M5, M5 Pro, or M5 Max Mac, Parallels Desktop lets Windows 11 on Arm run in a virtual machine alongside macOS. You can use Windows applications without rebooting your Mac and keep them available alongside your Mac applications, files, and other tools.

For most M5 users, Parallels Desktop handles CPU and memory allocation through its Automatic settings. Having an M5 Mac with more memory doesn't mean you need to manually configure the VM before running Windows.

If your workload needs more specific resource allocation, you can adjust CPU and memory settings for individual virtual machines. Coherence mode can also let you use Windows applications alongside Mac applications without keeping the traditional Windows desktop in view.

For installation and setup details, see Running Windows on Apple M5 Macs with Parallels Desktop.

Already have an M5 Mac and want to see how your Windows workload behaves on it?

Try Parallels Desktop free for 14 days.