- The Nvidia N1X chip was spotted in a geekbench result
- The specifications show that the GPU integrated in 6,144 cuda coda
- This is equivalent to the RTX 5070 for the number of pure nuclei, but there is much more to take into account when it comes to performance
Remember the rumor processor of Nvidia which caused a Buzz on the vine last year? We have apparently seen this consumption chip in a reference leak, where the details of the overturned specifications are the key aspect.
Tom’s material reports that the N1X chip, which is based on ARM (like Snapdragon X processors from Qualcomm), has been identified in a Geekbench result, in particular for the OpenCl test (graphics), where it obtained a score of 46,361.
This score is almost insignificant at this stage. This is an early engineering sample of N1X (in theory), and even then, if you want to assess graphic performance, Geekbench is far from being the first choice of synthetic references.
As indicated, however, this gives us an enticing overview of the specification, which shows that (add salt now), the N1X will have 20 cores, apparently divided into a pair of clusters with 10 hearts. It is the processor itself, but we also see the GPU integrated here, which is shown that it has 48 multi -processes in streaming – which is equivalent to 6,144 Cuda nuclei.
It looks a lot like, right? Well, this is the case, and in fact, those who know Nvidia’s graphics cards will realize that it is in the base stage of a GPU of the current mid -range generation – to be precise, the RTX 5070, which, in fact, has this number of exact nucleus.
Analysis: prudently optimistic
So, do we get a compact consumption chip that could go to budget laptops or pocket computers to offer the same image frequencies as the powerful RTX 5070? In a word, no, but the N1X always seems to be put in place as a promising piece of silicon, and the one who will have competitors sit and take note.
As for the reasons why performance cannot simply be taken from the number of nuclei observed on the GPU here – this is not a corrective on the RTX 5070 in this reference, of course – there are other important factors in play outside the basic base.
This includes the clock speed and the power supply to the GPU, which is a very different scenario with graphics integrated into a chip like this compared to a complete graphics card in a desktop PC. In addition to considering the power envelope, also throw the bandwidth limitations – in terms of piping tasks in the memory of the system, without an embedded of of course – and the result is a good part of opposite winds.
This will not prevent the N1X from being an artist potentially sterling for an all-in-one chip, but there is not much points trying to guess at the exact level of the graphic performance he could provide at this stage. (Certainly not the reference index disclosed here, as already indicated).
Tom’s makes an interesting observation, which is that the disclosed specifications correspond to the “Superchip” GB 10 of NVIDIA built for powerful AI performance and by inaugurating the era of the tiny IA supercomputer (illustrated above). There is no reason why Nvidia can not go on another turn on this subject for devices targeted by the consumer, including mini PCs and laptops, and players are particularly enthusiastic about possible use in pocket computers.
For the moment, however, this is still very good in rumors. If we believe the previous speculation, we could see the consumer processor of Nvidia later revealed this year, before a launch at the beginning of 2026.
