NVIDIA is planning a frontal assault on the PC processor market, reports TechPowerUP ‘s btarunr, and not sneaking in through the back door, but with a full door-to-door mentality. TechPowerUp reports on a Geekbench leak of the upcoming NVIDIA N1X, a SoC that comes with 20 ARM cores and an integrated GPU with a whopping 48 Blackwell SMs – equivalent to the performance level of an RTX 5070, but on-die, without a dedicated card.

Geekbench 6.4.0: First traces in the lab
According to the entry in the Geekbench benchmark, the N1X identifies itself with:
- 20 CPU cores
- 48 GPU compute units (OpenCL)
- 46.361 points in the OpenCL test
That’s an announcement – not only for mobile devices, but also for ultrabooks, tablets and possibly even “real” PCs. Apple’s M3 Pro/Max and AMD’s Ryzen “Strix Halo” in particular are likely to be looking nervously at the data sheet.
Heterogeneity: Big-LITTLE on steroids
Although it is not yet clear exactly how the 20 cores are divided up, everything points to an asymmetrical design à la Arm DynamIQ – in other words: performance cores for full load and efficiency cores for idle optimization. The strategy is familiar from Apple, Qualcomm and MediaTek – and NVIDIA now also wants a slice of the Windows 11 pie. Or rather: from the upcoming “Copilot AI PC” ecosystem that Microsoft is currently establishing with vigor. The motivation is clear: Windows 11 on ARM is no longer the joke product of 2018, but is mutating into a new playground thanks to NPU focus and AI offload – and NVIDIA does not want to leave this field to Qualcomm, Apple or MediaTek without a fight.

Blackwell inside: 48 SMs put the pressure on
The highlight of the N1X: the integrated GPU is based on the latest Blackwell architecture. And this is where things get dicey for the competition: 48 SMs – that corresponds exactly to the desktop version of the GeForce RTX 5070. Of course, the clock rate will be lower and the power limit will be more restrictive – but in terms of shader density alone, this iGPU monster is miles ahead of anything Apple, AMD or Intel have ever offered in the SoC segment. In other words, the N1X could be the fastest integrated GPU on the market at launch – provided that thermal and energy conditions allow this in practice.
The strategic context: Apple, Qualcomm, AMD in our sights
NVIDIA is sniffing the morning air. With the AI hype as a tailwind and Microsoft’s opening of the ARM architecture to third-party manufacturers, the company now wants to go where it has never been before: onto the CPU desktop. The last time NVIDIA designed its own PC processor was with the Tegra K1 – successful in theory, but ignored by the market.
This time the game is different:
- Apple has proven with M1-M3 that ARM works on the desktop.
- Microsoft wants to push its own NPU-based platforms – and is opening the door.
- Qualcomm is warming up with the X Elite.
- AMD is coming with Strix Halo and XDNA NPUs.
NVIDIA does not want to be last. Instead, they are bringing the power of the Blackwell GPU to an all-rounder chip – with the intention of helping to shape the ecosystem in the long term.
The N1X is not a gimmick – but NVIDIA’s desktop debut with a sharp blade
20 ARM cores, 48 SMs – this is not a trial balloon, but a real market attack. If NVIDIA manages to bring this chip into mass production efficiently, scalably and with suitable OEM partners, then the N1X could become an arm-based counterweight to Apple Silicon and AMD’s Ryzen AI.
An attack on several fronts:
- Against Apple with performance-per-watt.
- Against AMD with GPU density.
- Against Qualcomm with CUDA ecosystem and machine learning power.
The only question that remains is: How expensive will it all be? And: Will Microsoft let NVIDIA onto the Copilot platform – or will the bouncer Qualcomm remain alone in the VIP zone for a while longer?
Source: TechPowerUP, Geekbench

































7 Antworten
Kommentar
Lade neue Kommentare
Mitglied
Urgestein
Urgestein
Urgestein
Neuling
Mitglied
Mitglied
Alle Kommentare lesen unter igor´sLAB Community →