The Apple M1 GPU is an integrated graphics card with 8 cores (one of which is deactivated in the entry-level MacBook Air) built by Apple and included in the Apple M1 SoC. According to Apple, it is speedier and uses less energy than competitor goods (like the Tiger Lake Xe GPU).
Apple’s new M1 Ultra is designed to outperform Nvidia’s RTX 3090.
Last year, Apple’s M1 Pro and M1 Max chips served as a GPU-shaped warning to Nvidia and AMD, and now Apple has turned up the heat with its new M1 Ultra.
Apple claims it will compete with Nvidia’s massive RTX 3090 graphics card, which is currently the fastest GPU on the market.
But how does Apple believe it can outperform an RTX 3090? It turns out that last year’s M1 Max contains some sort of special sauce.
Apple’s new M1 Ultra is an unexpected mix of two M1 Max dies fused to form a single strong chip. Apple can combine two chips into one using the M1 Max’s secret high-speed interface.
As a result, the M1 Ultra processor has twice as many CPU cores, twice as much memory, twice as much memory bandwidth, and, most crucially, twice as many GPU cores.
This combination is known as UltraFusion by Apple, and it is effectively Apple’s own 2.5D chip packaging approach.
For years, the semiconductor industry has used chiplets to construct CPUs, with AMD’s Zen 2 and Zen 3-based Ryzen chips leading the field in terms of current chipset designs and, until recently, performance.
Apple competitors Intel, Samsung, and Qualcomm are collaborating on a new standard that might allow firms to construct computers out of Lego-like chiplets. Apple is ahead of the competition with a processor that combines two different GPUs.
Nvidia and AMD have already developed similar ways for integrating two GPUs, but as AnandTech points out, it appears that Apple has achieved the holy grail of multi-GPU design here.
Apple’s UltraFusion technology allows for a staggering 2.5TB/s of bandwidth between the two M1 Max CPUs.
That’s a significant increase in bandwidth over Nvidia’s NVLink for SLI or AMD’s Infinity Fabric, both of which are used as high-speed communications between GPUs.
Because of Apple’s high-speed connectivity, the two discrete M1 Max GPUs will appear in macOS as one GPU, allowing apps to readily tap into the combined capability.
That should imply that apps and games won’t have to do anything special to take advantage of the M1 Ultra’s power, whereas in the past, games had to natively support Nvidia’s SLI implementation on Windows to experience any performance improvements.
With the RTX 30-series, Nvidia has effectively abandoned multi-GPU functionality, with only the RTX 3090 enabling NVLink support.
However, using two RTX 3090s for work or gaming rigs yields very varied results. Games that natively support SLI will provide performance improvements, whilst the majority will not, and some will even suffer a performance reduction.
The RTX 3090 is the fastest GPU on the market right now until Nvidia’s long-delayed RTX 3090 Ti arrives and Apple claims the M1 Ultra can outperform a single RTX 3090 while drawing 200 watts less power.
Last year, Apple claimed that the M1 Max outperformed the RTX 3080, but real-world results were mixed.
When it came to productivity-focused workloads, the M1 Max outperformed the RTX 3080. For Adobe Premiere Pro jobs, several reviewers considered the M1 Max to be slightly slower than a comparable RTX 3080 machine, however, this relied on the task at hand.
Apple has crammed its M1 Ultra inside the new Mac Studio, a desktop computer that is only slightly larger in volume than Nvidia’s RTX 3090. Because of the M1 Ultra, the Mac Studio is extraordinarily powerful, and Apple has planned it to replace the 27-inch iMac and even Mac Pro models for many people.
Apple’s M1 Ultra and Mac Studio are aimed squarely at productivity applications rather than games.
In gaming tests, both the M1 Max and M1 Pro performed comparably to an RTX 3060 in numerous titles.
The M1 Ultra will not magically alleviate the shortage of macOS games or the fact that the majority of cross-platform games are still x86.
We’ll have to wait for reviews to see what doubling the GPU cores does for productivity app performance, but Apple’s already great start with the M1 Pro and M1 Max appears to be accelerating with the M1 Ultra.
It took Qualcomm and Microsoft years to give a laptop-like performance on ARM-based CPUs for Windows, and Apple is now delivering workstation-level performance in this area.
The Mac Pro is where Apple’s processor design will get intriguing. Apple’s current Mac Pro is powered by Intel Xeon CPUs and Radeon PRO W6000X GPUs from AMD.
Yesterday, Apple concluded its event by teasing an Apple Silicon version of its Mac Pro machine “for another day.” Last year, Bloomberg reporter Mark Gurman disclosed that the Mac Pro will have up to 40 CPU cores and 128 GPU cores. If you’re keeping track, that’s two M1 Ultras.
Final thought
The M1 Ultra inside the Mac Studio will show us how effectively Apple’s core count and GPU performance scale up to high-end computers. It’s a taste of what’s to come when Apple equips the Mac Pro with the best of what Apple Silicon has to offer.
Related Article:
How Do You Fix the Constant GTA5 PC Crash in Game? (Explained)