NVIDIA's long-awaited "full-power" SoC chip has been the subject of successive leaks.
For many months now, MediaTek and NVIDIA have been keeping us waiting with their new chip, designed to compete with "just about everyone else" in the market for microprocessors with a solid integrated graphics solution. It's the joint efforts of the two companies that should make it possible to combine the best that the electronics industry has to offer: MediaTek for know-how in smartphone chips and energy efficiency, and NVIDIA for - need we say it again - everything to do with graphics performance and artificial intelligence.
Two chips appear to be under development, and the most powerful is known as the N1x, which may not be its final commercial name. To be seen. In any case, one thing is certain: MediaTek and NVIDIA are pinning their hopes on this chip, which is worrying many of the industry's regulars, whether AMD, Intel or Qualcomm. The N1x is what we call a SoC for System on a Chip, and should therefore integrate the functions of CPU, GPU, NPU, I/O management and RAM into a single chip. On this last point, we don't yet know much, but successive leaks from the Geekbench databases, notably relayed by VideoCardz, point to a solution with at least 64 GB of RAM.
The ultimate NVIDIA N1x leak © VideoCardz
Another particularly useful piece of information is that we're talking about 48 graphics processing units, an unusual designation for NVIDIA, which generally prefers to talk in terms of CUDA cores. If we follow the brand's usual practice, 48 compute units should result in a whopping 6144 CUDA cores, a far from ridiculous figure for an integrated graphics solution. That's quite a lot more than an RTX 5070 Laptop graphics card (4608 cores), which also boasts a TDP of just 100 watts. Other rumors point to a possible TDP of 170 watts for the N1x, which, if true, would undoubtedly make it a real graphics powerhouse. For a variety of technical, logistical and software reasons, NVIDIA does not intend to launch the N1x before next year.
