M1 Ultra VS RTX 3090: Apple’s new chip gets smoked by Nvidia
No, an ARM chip isn't better than the most powerful GPU on the market.
WePC is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Prices subject to change. Learn more
When Apple revealed their new M1 Ultra chip at their Peek Performance event, they unveiled the Mac Studio, and with the verdict out on the device itself, critics have put to the test Apple’s claims that it will beat out the highest-end discrete GPUs. The Apple M1 Ultra merges two M1 Max chips together, a little bit like an SLI bridge to combine their performance and enable them to reach workload performance that has never been seen before on an ARM chip. These claims remain true for many of the M1 Ultra’s claims, and there is indeed something to be said about the proliferation of ARM silicon among Apple’s chipsets, but it’s going to be a good while before it can catch up with the sheer horsepower under the hood that Nvidia’s most powerful GPU can provide (Until the 3090Ti releases, of course.)
What claims did Apple make about the M1 Ultra’s performance?
Apple’s graphs in their product announcements are always pretty vague, and you only need to look at the Y-axis on this graph in order to understand how frankly silly Apple’s claims can come across as. ‘Relative’ performance is such as strange barometer to judge your product by, and while indeed, the M1 Ultra might be able to match the RTX 3090 in terms of GPU performance in certain workloads, reviews are coming out that refute this claim entirely. The Verge reports that the RTX 3090 outperformed the M1 Ultra by over 100% in Geekbench 5 compute testing. The M1 Ultra scored 83,121 in their testing, while the RTX 3090 machine managed 215,034.
So, where is this relative performance that Apple claimed that the M1 Ultra claimed to match? They appear to be pretty much nonexistent. Claims like this are nothing new for Apple, and while their acceleration of ARM processors across their entire product stack has been impressive, and their gains in performance-per-watt has also been equally respectable, it’s looking like it’s going to take a little bit more time before Apple manages to catch up on the pure performance front to match the biggest and best graphics cards in the industry.
The market is becoming saturated at the higher-end, with Nvidia, AMD and now Intel entering the fray. Though, Apple has yet to match up in terms of pure GPU performance than any of these brands, if this testing holds true.
Apple M1 Ultra VS RTX 3090
The winner in the battle of the Apple M1 Ultra VS the RTX 3090 is clearly the 3090, it’s not even a close fight. The sheer farce of Apple even claiming that the M1 Ultra reaches near-RTX 3090 level performance is false, as per reports via critics. We’re not even sure that the M1 Ultra would even stack up against an RTX 3060, forget about the 3090. Geekbench once again reports that an RTX 2060 Super outperforms the M1 Ultra, so here’s a fun little graph to poke your head at, thanks to Geekbench.
|RTX 2060 Super
|Apple M1 Ultra
As you can see, Apple’s claims were so far off the mark of being better than the ‘highest-end’ discrete GPU that it’s outperformed by a midrange graphics card that’s almost three years old. When looking at the relative performance of the card against Nvidia’s current lineup of Ampere graphics cards, it does not even become a conversation. This is why the ‘relative performance’ of Apple’s graphs can be so misleading.
It’s possible that Apple decided that relative performance meant performance-per-watt, but it’s still a misleading moniker to lead within marketing materials. This is something that is actively anti-consumer, but Apple’s no stranger to doing that either if it makes their machines look good. We’re not saying that Apple’s M1 Ultra chip is bad, it’s far from it, and provides huge benefits for those looking for oodles of power in a small package, but instead, the reason why this misleading marketing is so abrasive is that a consumer could parse this information as being applied to different workloads, the ‘relative’ part of their graph means that they can easily get away with this scot-free, by simply not specifying why the M1 Ultra is so good. But, by this point, the entire industry knows that ARM chips are more efficient when it comes to energy usage, so the entire graph can be construed as being invalid. This frankly terrible usage of data is one of the things that are just designed to make Apple look good when they reveal a new product, and not give consumers any information that they want to know about the actual power under the hood.
The other companies in the tech space are no strangers to this, either, but this is a fairly egregious claim that Apple has made in the announcement of the product itself, so cool your jets Cupertino, you’re not quite that powerful yet.
3D artists and gamers should stick with dedicated GPUs
Apple’s claims and tendrils in the creative industry might look attractive to most, but the reality as shown in the data is that the M1 Ultra and RTX 3090 are just in completely different leagues, for those who work in 3D art or are into hardcore gaming, then your bets are still firmly placed in the realm of ensuring you have a good dedicated GPU to get the most performance, and more importantly, to save you time while working.
Sure, Geekbench is a single benchmark, and Apple has made improvements elsewhere, but when it comes to the raw power of the chip, sometimes the more traditional solutions are often the best. It’s going to be a good while before Apple manages to catch up with the pure power of other chips, so instead of looking at the Mac Studio as a desktop replacement, think of it instead as a viable complement to your existing powerful workhorse machine. Thinking of it as anything more than that is just going to end up wasting time, and money.