NVIDIA RTX Spark introduces unified CPU and GPU architecture

NVIDIA has introduced the RTX Spark platform, designed to address the performance bottlenecks between traditional CPUs and GPUs. The platform utilizes the N1X architecture, which unifies an ARM-based CPU with an NVIDIA Blackwell GPU within a single package to facilitate local AI execution, content creation, and high-end gaming.

The technology is available in two primary configurations:

- The N1X 18-core version, featuring 9 performance and 9 efficiency cores, alongside 120 CUDA cores. - The N1X 20-core version, featuring 10 performance and 10 efficiency cores, alongside 144 CUDA cores and 321 AI TOPS of computing power.

This shift toward unified architecture aims to support local AI agents that do not require external servers, while maintaining gaming performance through technologies like DLSS 5.