Nvidia’s long-awaited Arm-based chip for PCs reportedly spotted running Geekbench very badly

The numbers looks horrible but it doesn't mean much, for now.

The numbers looks horrible but it doesn't mean much, for now.

Back in November we reported that Nvidia was planning to put an ARM CPU for the PC into production in 2025. Now it seems the chip, or at least some version of it, may have been spotted crunching very slowly through Geekbench.

The chip on Geekbench is registered as the N1X, which aligns with rumours from January that Nvidia would unleash a “high end” N1X Arm-based chip this year, with a more mainstream N1 variant following in 2026. Both chips are said to be built on TSMC’s N3 node and engineered in partnership with Mediatek.

Ostensibly, the scores for the new Nvidia chip look terrible (via X user Jukanlosreve). It racks up just 1,169 single-core points in Geekbench 6. That compares with 3,831 for Apple’s current M4 chip, which is also Arm based.

The multi-core score is even more feeble at just 2,417 points to the M4’s 15,044. And that’s just the basic M4 chip. MacBooks with the M4 Max chip can breach 25,000 points, or over 10 times the performance.

Notably, the Nvidia N1X listing on Geekbench’s results page indicates a four-core CPU. It’s almost certain that any “high end” Nvidia ARM CPU will have more than four cores. Indeed, the mainstream N1 variant will also very likely have more than four cores.

In other words, whatever it is that has popped up in Geekbench, it’s extremely unlikely to represent anything close to the final Nvidia N1X chip. That also applies to the chip’s clockspeed and firmware.

It’s listed on Geekbench at 3.2 GHz, which is decent for an Arm CPU. But there’s no way on knowing if it was really running at that speed or if the chip was fully functional in other ways.

Long story short then, the interesting thing here isn’t anything to do with the purported specs of four cores and 3.2 GHz, both of which won’t be final, but the simple fact that a chip called N1X is up and running somewhere and returning any kind of benchmark results at all.

The idea of an Arm-powered PC, of course, is nothing new. Qualcomm is already having a serious tilt at it with the Snapdragon X series of chips, while Apple has proven with its M-series hardware that the Arm instruction set can go toe-to-toe with any x86 CPU. Indeed, in terms of IPC, Apple’s chips are currently miles ahead of Intel and AMD.

The problem, for us at least, has always been game support. Of just about any application type, games represent the hardest challenge for a whole new CPU instruction set.

Gaming on the Snapdragon X via emulation has proven patchy thus far. However, if any company can get PC gaming working on Arm, it’ll be Nvidia. That could be through an incredibly powerful CPU that can emulate seriously fast, convincing game devs to code native versions of their games, or maybe a combination of both.

Metro Exodus running on a Snapdragon X Elite laptop

We’ve tried gaming on Arm and it’s, er, OK? Ish? OK, kinda not. (Image credit: Future)

Exactly what the involvement with Mediatek will bring isn’t clear, meanwhile. It’s thought one important driver for the Mediatek tie-in may be 5G cellular networking support which could be critical for the enterprise version of the new chip.

Consumer PCs arguably don’t need 5G and, frankly, I’d rather see Nvidia designing the CPU cores than Mediatek. So, here’s hoping the Mediatek bit is indeed limited to just the wireless comms.

Your next upgrade

Nvidia RTX 4070 and RTX 3080 Founders Edition graphics cards

(Image credit: Future)

Best CPU for gaming: The top chips from Intel and AMD.
Best gaming motherboard: The right boards.
Best graphics card: Your perfect pixel-pusher awaits.
Best SSD for gaming: Get into the game ahead of the rest.

For now, no other details are known. For instance, which graphics architecture will the new APU use? It’s all speculation for now, but the claim that N1X and N1 are on TSMC’s N3 node suggest that the chip could use Nvidia’s next-gen Rubin GPU technology rather than the current Blackwell architecture.

That’s because GPU architectures are somewhat tied to process nodes. Blackwell is built on N4 while Rubin is expected to be on N3. Now, it’s not impossible that Nvidia has ported Blackwell to N3 for the N1X chip. But that would require a full redesign for the new node, which costs time and money.

Arguably, it could be easier to go with an architecture that’s already been built for TSMC’s N3 design rules. And that would be Rubin.

Ultimately, time will tell. But a true “high end” Arm chip from Nvidia with next-gen graphics would certainly be exciting. We can’t wait to see what Nvidia has cooking.

About Post Author

Leave a Reply

Your email address will not be published. Required fields are marked *