Benchmarked: Nvidia's RTX Spark notebook delivers serious local AI compute, at a price

Nvidia's RTX Spark notebook system is now available, boasting a powerful CPU, lightning-fast integrated graphics, and up to 128 GB of unified memory in a sleek package. However, the high price point starts at around $2,600 and can exceed $6,000 for the top models. Despite the steep cost, the RTX Spark is ideal for AI enthusiasts, researchers, and creative professionals who require large amounts of video memory and prefer Nvidia's CUDA ecosystem over AMD's ROCm.

The Register tested the OmniBook Ultra, HP's customized version of the RTX Spark platform, and found it to be one of the fastest Windows notebooks for local AI. The notebook features a stunning OLED touchscreen, a stiff hinge mechanism, a traditional keyboard, and excellent speakers, although it lacks a full-size SD card slot. The standout feature is the N1X chip, which features a MediaTek-derived CPU chiplet, Nvidia's NVLink Fusion chip-to-chip interconnect, and a Blackwell GPU.

The N1X comes in two versions, one with 18 CPU cores and a GPU with 5,120 CUDAs, and another with up to 20 CPU cores and 6,144 CUDA cores. While the chip itself is not entirely new, it does offer improvements such as higher memory bandwidth, translating to faster token generation in AI inference and agentic workloads.