Apple’s charts have the M1 Ultra ready for an RTX 3090 battle it could never win

When Apple introduced the M1 Ultra — the company’s most powerful internal processor to date and the crown jewel of its brand-new Mac Studio — it did so with graphics bragging that the Ultra was capable of beating Intel’s best processor or Nvidia’s RTX 3090 GPU all. its own. The charts, in Apple’s recent fashion, were maddeningly labeled with “relative performance” on the Y-axis, and Apple doesn’t tell us what specific tests it runs to arrive at whatever numbers it uses to then calculate “relative performance.” calculate.

But now that we have a Mac Studio, we can say that in most tests, the M1 Ultra is not actually faster than an RTX 3090, as much as Apple would like to say it is.

To hear Apple tell it, the M1 Ultra is a silicon marvel, one that combines the hardware of two M1 Max processors for a single chipset that’s nothing short of the “world’s most powerful chip for a personal computer.” And if you just look at Apple’s charts, you might be tempted to believe those claims.

On the map here, the M1 Ultra is doing beat the RTX 3090 system for “family member” GPU performance while consuming vastly less power. It’s a great achievement!

But that’s because, for lack of a better term, Apple’s chart has been cropped. The company only shows heads for the areas where the M1 Ultra and the RTX 3090 compete, and it’s true: in those circumstances, you get more bang for your buck with the M1 Ultra than you would on an RTX3090 .

But what the chart doesn’t show is that while the M1 Ultra’s line more or less stops there, the RTX 3090 is a lot more power it can use — just take a look at some of the benchmarks from The edge‘s review:

Image: The Verge

Image: The Verge

As you can see, the M1 Ultra is an impressive piece of silicon: it easily outperforms a nearly $14,000 Mac Pro or Apple’s most powerful laptop. But it seems that Apple just isn’t showing the full performance of the competitor it’s chasing here.

It’s like claiming that because at 80 miles per hour your electric car can use significantly less fuel than a Lamborghini, it has a better engine – without mentioning that a Lambo can still go twice as fast.

And yes, it’s very impressive that Apple achieves so much with (relatively) so little power. I’m sure Apple’s chart is accurate to show that the M1 Ultra outperforms the RTX 3090 in that particular comparison at the relative power and performance level. But it basically misses the rest of the chart where the 3090’s line shoots way past the M1 Ultra (though also with a lot more power).

It feels like the chart should probably look more like this:

The thing is, Apple didn’t need to do all these chicanes: the M1 Ultra is legitimately something to brag about, and the fact that Apple has managed to seamlessly merge two disparate chips on this scale into a single unit is an impressive feat whose the fruits apparently are in almost every test my colleague Monica Chin ran for her review.

Apple’s UltraFusion connection technology actually does what it says here, offering nearly double the M1 Max in benchmarks and performance tests. Apple taped two M1 Max chips together with duct tape and basically got the performance of twice the M1 Max. No other chipmaker has ever really accomplished this.

That’s fantastic and much more impressive and interesting for Apple to have spent time demonstrating its best, bleedingest edge chip beating out aging Intel processors from computers that plotted the latest generations of chip design or tampered graphs that threaten the M1 Ultra. to fail under real-world scrutiny.

It’s okay that Apple’s latest chip can’t beat the world’s most powerful dedicated GPU! The 3090 on its own is nearly the size of an entire Mac Studio and costs nearly a third as much as Apple’s most powerful machine. But I can’t help but wish Apple would focus on accurately showing customers the M1 Ultra’s true strengths, advantages and triumphs rather than creating charts that have us chasing benchmarks that — deeply Inside – Apple needs to know it can’t match. At least not yet.