July 12, 2025
Trending News

What NVIDIA’s ARM SoC will be for the next generation of Windows AI PCs

  • May 28, 2024
  • 0

NVIDIA ARM SoC for next generation IA PC with Windows, will be available in 2025as indicated by the head of Dell, Michael Dell, in an interview with Bloomberg

What NVIDIA’s ARM SoC will be for the next generation of Windows AI PCs

NVIDIA ARM SoC for next generation IA PC with Windows, will be available in 2025as indicated by the head of Dell, Michael Dell, in an interview with Bloomberg TV, where the CEO of the green giant was also present.

For months there were rumors about the development of the project in which NVIDIA will introduce ARM processors for various purposes. One will be the one for consoles, which the Nintendo Switch 2 would certainly mount, the second is mobility for tablets and smartphones, and the third, and the most important, considering the madness we are in with AI: personal computers. the Windows platform on ARM.

Last week we gave him a big helping hand with everything that was announced for Microsoft’s Copilot+ PC. It seems clear Microsoft won’t beat Apple with Qualcomm alone and vendors the size of NVIDIA will go a long way in reducing Cupertino from their privileged position in PCs with this architecture.

What will NVIDIA’s ARM SoC be like?

NVIDIA is no stranger to SoCs, and its Tegra series has sold millions of units over the past decade, largely thanks to sales of the Nintendo Switch console. But a new ARM SoC would be a leap forward.

From what we know so far, NVIDIA will be combining new ARM CPU cores, Cortex-X5 codenamed ‘Blackhawk’ and LPDDR6 memory, also new, which we had news about last week. Of course, the graphics technologies would be their own and cutting-edge, as they would correspond to the next generation ‘Blackwell’. This iGPU would provide the NVIDIA ARM SoC with cutting-edge graphics acceleration capabilities, including DirectX 12 Ultimate and Vulkan.

NVIDIA ARM SoC

The responsibility of accelerating AI tasks would fall on Tensor cores. NVIDIA could simply expose the GPU as a Windows device virtual NPU and use a number of translation layers for TensorRT. Since Microsoft’s requirement for an NPU capable of limiting a local Copilot+ session is 45 TOPS, For NVIDIA, this would be a very easy-to-achieve brand. As for the production node, there is talk of N3P (3 nm processes) from the TSMC foundry.

There is no doubt that NVIDIA’s ARM SoC could be a turning point in Windows on the ARM platform. And a big competitor for Qualcomm, Apple and also Intel, whose x86 is being attacked from multiple fronts. The industry is moving…

Source: Muy Computer

Leave a Reply

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