• Nvidia’s N1X chip has been noticed in a Geekbench outcome
• The specs present the built-in GPU has 6,144 CUDA cores
• That equals the RTX 5070 for pure core rely, however there’s far more to think about in terms of efficiency

Bear in mind Nvidia’s rumored CPU that brought on fairly a buzz on the grapevine final yr? We have apparently now seen this client chip in a benchmark leak, the place the spilled spec particulars are the important thing facet.

Tom’s {Hardware} reviews that the N1X chip, which is Arm-based (like Qualcomm’s Snapdragon X CPUs), has been noticed in a Geekbench outcome, particularly for the OpenCL (graphics) take a look at, the place it scored 46,361.

That rating is just about meaningless at this level. That is an early engineering pattern of the N1X (in concept), and even then, if you wish to gauge graphics efficiency, Geekbench is way from the primary selection of artificial benchmarks.

As famous, although, this provides us a tantalizing glimpse of the spec, which reveals that (add salt now) the N1X could have 20 cores, apparently break up right into a pair of 10-core clusters. That is the processor itself, however we additionally see the built-in GPU right here, which is proven to have 48 Streaming Multiprocessors – that equates to six,144 CUDA cores.

That feels like quite a bit, proper? Nicely, it’s, and actually, these acquainted with Nvidia’s graphics playing cards will understand that that is within the ballpark for a mid-range current-gen GPU – to be exact, the RTX 5070, which, in reality, has that actual core rely.

Evaluation: cautiously optimistic

So, are we getting a compact client chip that would go in funds laptops or handhelds to ship the identical body charges because the mighty RTX 5070? In a phrase, no, however the N1X nonetheless seems to be shaping up as a promising piece of silicon, and one that may have rivals sitting up and taking discover.

As to the explanation why efficiency cannot merely be drawn from the variety of cores seen on the GPU right here – it isn’t a patch on the RTX 5070 on this benchmark, after all – there are different essential components at play apart from the fundamental core rely.

That features the clock pace and the facility equipped to the GPU, which is a really totally different situation with built-in graphics in a chip like this versus a full-on graphics card in a desktop PC. In addition to contemplating the facility envelope, throw in bandwidth limitations too – by way of piping duties over to the system reminiscence, with no on-board VRAM after all – and the upshot is a great deal of headwinds.