- Nvidias n1x -chip has been seen in a geekbonch -result
- The specifications show that the integrated GPU has 6,144 CUDA -Kernes
- It corresponds to RTX 5070 for pure core numbers, but there is much more to factor in when it comes to performance
Do you remember Nvidia’s rumored CPU who caused a pretty hum on grapevine last year? Apparently, we have now seen this consumer chip in a benchmark leak where they wasted spec -details are the central aspect.
Toms Hardware reports that the N1X chip, which is arm-based (such as Qualcomms Snapdragon X CPUs), has been seen in a geekbench result, specifically for the OpenCL (graphics) test where it scored 46,361.
This score is largely meaningless at this time. This is an early technical test of N1X (in theory), and even then, if you want to measure graphics benefit, Geekbench is far from the first choice of synthetic benchmarks.
As noted, however, this gives a tempting glimpse of spec, showing that (adding salt now) N1X will have 20 kernels, apparently divided into a few 10-core clusters. It’s the processor itself, but we also see the integrated GPU here, which turns out to have 48 streaming multiprocessors – equivalent to 6,144 cuda kernels.
Sounds a lot, right? Well, it is, and in fact, those who are familiar with Nvidia’s graphics card will realize that this is in the ballpark of a mid-range power-gen gpu-to be precise, RTX 5070, which actually has the exact core count.
Analysis: Gently optimistic
Then we get a compact consumer chip that can go in the budget laptops or handhelds to deliver the same image speeds as the mighty RTX 5070? With a word, no, but N1X still seems to be shaped like a promising piece of silicon, and one that wants rivals sitting up and noticing.
As for the reasons why performance can not only be drawn from the number of cores seen on GPU here – it is not a patch on the RTX 5070 in this benchmark, there are of course other important factors played except for the basic core number.
It includes the watch velocity and power delivered to GPU, which is a very different scenario with integrated graphics in a chip like this versus a full graphics card on a desktop PC. In addition to considering the power convoluted, you must throw bandwidth restrictions as well – with regard to pipe tasks for the system memory without of course on board VRAM – and that is a good part of headwinds.
It does not prevent N1X from being a potential sterling artist for an all-in-one chip, but there is not much point trying to guess at the exact level of graphics performance it may give at this time. (Certainly not from the leaked benchmark here, as already mentioned).
Tom’s makes an interesting observation, which is that the leaked specifications match Nvidia’s GB10 ‘Superchip’ built for powerful AI performance and initiated the era with the small AI super computer (pictured above). There is no reason why Nvidia could not put another spin on this to consumer-targeted devices, including mini-PCs and laptops, and in fact players are particularly excited about the possible use in handhelds.
For now, however, this is still very much in the rumor. If there is to be believed earlier speculation, we can see that Nvidia’s consumer -CPU was revealed later in the year, prior to a launch at the beginning of 2026.
