It’s about time Digital Foundry took a look at Apple Silicon, and today we’re looking at the high-end chips in the lineup. We’re focusing on the monster that is the M1 Ultra, which is in the latest Apple Mac Studio, but we’ll also take a look at the M1 Max in the MacBook Pro. However, it’s the M1 Ultra that really grabs our attention: this system-on-a-chip represents the highest-end computer processor Apple has designed yet, and the company claims it should be as fast as a desktop of high-end Windows. With 20 CPU cores, a 21 teraflop GPU and 800GB/s of memory bandwidth, it looks like it could be, but how does it measure up in real-world tests and perform?
The M1 and M2 lines are the culmination of a long journey that has seen the firm transition from third-party design-based SoCs and IP, moving all chip design in-house as realistically as possible. Apple designs its own GPUs, its own CPUs, and handles SoC design and integration. This translates into enormous control over processor design, the kind of control you’ll need to scale a phone processor for high-end desktops.
Which brings us to the M1 Ultra. Since 2020, Apple has been moving its Mac desktops and laptops away from Intel CPUs and AMD GPUs and toward its in-house SOCs, taking the same core technology from iPhones and integrating it into PCs. Apple started out with lower-end, lower-powered form factors, but finally made its way into high-end desktops with the launch of the Ultra a few months ago. However, the M1 Ultra isn’t really its own unique chip. It’s actually two M1 Max SoCs connected via a 2.5TB/s high-bandwidth interposer. To the OS and user it looks like a monolithic chip with 1 CPU and GPU, but it’s actually two chips connected via a single interconnect with the performance to support a dual-chip GPU and CPU.
Digital Foundry video analysis of the M1 Ultra and M1 Max processors, compared to powerful PC equivalents such as the Core i9 12900K and RTX 3090.
The Ultra features 20 CPU cores, split between 16 performance cores and 4 efficiency cores in a configuration similar to modern Intel designs. Although clock speeds may be lower than desktop PC CPUs, instructions per clock are higher on performance cores, leading to similar overall performance per core. At its highest spec, the 21-teraflop GPU packs 64 of Apple’s internal graphics cores, with performance similar to an RTX 3090 according to Apple, though we’ll get to that later. To top things off, the system comes with an impressive 800GB/s memory bandwidth to keep those GPU and CPU cores well fueled.
The M1 Ultra is currently only available on the Mac Studio desktop computer, which we tested in its maximum configuration, with 128GB of memory and an 8TB SSD. The most interesting thing is that this computer has a volume of only 3.7 liters, which is really small and only slightly larger than an Xbox Series S. It uses two blower-style fans that draw air through a large copper heatsink to dissipate the roughly 200W the system draws on load, which is a tiny fraction of the power used by a high-end desktop computer.
So let’s go ahead and actually measure the speed of this machine. We’ll start with gaming tests before closing with productivity and synthetic benchmarks. Is this machine really as fast as a high-end desktop PC, or maybe even faster? Let’s take a look at our gaming benchmarks, calculated using video capture the Digital Foundry way. Although internal benchmarks are very accurate these days, our philosophy is that the only frames that matter are the frames that actually make it to the hardware video output.
M1 Max (MBP 2021) M1 Ultra (Mac Studio 2022) RTX 3080M 150W (MSI GP66 Laptop) RTX 3090 (Desktop) Shadow of the Tomb Raider 31.0 49.0 29.0 65.0 Metro Exodus Total . of Warcraft 18.4 36.2 32.9 81.6 Wildlife Extreme Bench 20215 35498 24247 42451
It’s not a particularly big table because, unfortunately, there aren’t many high-end Mac games that we can test, especially when it comes to big-budget games. But here we have some titles, and the results are intriguing. For our gaming tests, we have a 16-inch Macbook Pro with the M1 Max chip fully enabled, our Mac Studio maxed out, an MSI GP66 gaming laptop, with an 11th-gen i9 and a mobile processor 150W RTX 3080 and a high-end desktop PC with a Core i9 12900K paired with the powerful RTX 3090.
Watching Shadow of the Tomb Raider. This isn’t a native Apple Silicon game, as the title was written for x86, so the M1 chips here have to use the Rosetta 2 translation layer to work, but it doesn’t really seem to have much impact on performance. . The benchmark sequence running at maximum settings at 4K shows the 3080M and the M1 Max neck and neck, while the Ultra sits between the M1 Max and the 3090. The Ultra is a solid performer and a reasonable scale from the Max, but it isn’t. We don’t hold the line against ultra-high-end GPUs.
Metro Exodus, the original non-RT version, has a decent Mac port, although again it was written for x86. The Ultra also splits the PCs here, while the Max does a good job of fending off the 3080M. On the other hand, there seem to be very serious issues with frame times and stuttering when vsync is disabled on the Mac for some reason, which I’ve noticed in these tests. Total War: Warhammer 3 is another x86 game, but it doesn’t seem to hold up as well as Metro or Tomb Raider. M1 Ultra lags far behind the 3090 here and barely keeps up with a high-end gaming laptop. Perhaps this can be attributed to a suboptimal port or issues with Rosetta’s translation.
Apple Silicon games and benchmarks are hard to find, but 3DMark’s Wildlife Extreme is a native app.
But what about Apple Silicon’s native games? There are very few games for Apple Silicon, and most of them are iOS ports, not mainstream PC software. However, there is one standout game that we can try out across platforms: World of Warcraft. It’s a full Apple Silicon version of Blizzard’s long-running MMO, but even though it runs natively, the same pattern appears with the M1 Ultra again falling squarely between the two PC systems, falling well short of the 3090, but still delivering performance in line with a high-end PC GPU. The Max can’t play at the limit while the 3080M hovers around 30fps. All of these systems would be perfectly fine with the game at remotely reasonable settings, of course – we’re running the game at essentially 8K internal resolution to create a proper stress test.
There’s a cross-platform gaming graphics benchmark that also runs natively on Apple Silicon: 3DMark Wildlife Extreme, which renders a set of relatively simple 3D scenes at 4K. Here, the Ultra falls a bit short of the 3090, but comes in 76 percent faster than the Max. Ultimately, the Ultra seems to fall somewhere below the 3090 in graphics performance, at least as far as we can tell from the cross-OS benchmarking. However, it’s still a powerful processor and seems to fit around the 3070 or 3080 level depending on the workload.
The scale of the M1 Max is reasonable, but not perfect. Typically, you should expect a 60-70 percent performance improvement over the single-chip option. Maybe the interposer is causing some minor hiccups here, since using multiple chips for one GPU requires a lot of bandwidth.
These results are really just for evaluating raw performance, as the Mac is not a good gaming platform. Very few games actually end up on the Mac, and the ports tend to be of low quality. If there is a future for Mac gaming, it will likely be defined by “borrowing” games from other platforms, either through wrappers like Wine or by running iOS titles natively, which they are capable of M1-based Macs. In the past, Macs could run games by installing Windows through Apple’s Bootcamp solution, but M1-based chips can’t natively boot into any flavor of Windows, not even Windows for ARM.
Blender (CPU Samples per min) M1 Max (MBP 2021) M1 Ultra (Mac Studio 2022) Core i9 10850K (Desktop) Core i9 12900K (Desktop) Monster 99.4 195.9 88.7 178 ,1 178.1 junkroom 18.4.3.3 3.185.3.3 18.3.18.3.1. CPU M1 Max (MBP 2021) M1 Ultra (Mac Studio 2022) Core i9 10850K (Desktop) Core i9 12900K (Desktop) Multi-core 12577 23580 9599 17446 Single-core 1774 1 2 187841 CPU Max. M1 Ultra (Mac Studio 2022) Core i9 10850K (Desktop) Core i9 12900K (Desktop) Multi-core 12259 23908 11171 25160 Single-core 1528 1531 1095 184KMax2020 Enke (Studio Max) ) Core i9 10850K (Computer desktop) Core i9 12900K (desktop) Time (mins:secs) 7:10 4:08 5:43 2:44
As you probably noticed from the table above, I also spent some time doing CPU comparisons on the M1 Ultra. I tested Blender, Geekbench, Cinebench, and Handbrake, and the Ultra’s results are convincing. Here we swapped the GP66 for my desktop PC, which includes a Core i9 10850K. Think of it as a Core i9 10900K with an almost perceptible clock speed reduction. In these tests, the 12900K and the M1 Ultra are very comparable. The two chips are essentially a match for multi-core performance, although the ultra-high frequencies the 12900K is capable of may give it an edge in some single-threaded tests. The 10850K and M1 Max are also very evenly matched.
The scale from M1 Max to M1 Ultra is almost linear in these runs, unlike our graphical benchmarks. On average, M1 Ultra is 88 percent faster, with some results approaching 100 percent. Linking two groups of cores through an interchip…