NVIDIA RTX Spark: Windows on Arm goes to a new level

The development of the Windows on Arm ecosystem reached a fundamentally new level. The hardware giant NVIDIA presented its first mobile system on a mass consumer computer crystal called RTX Spark. This powerful ARM chip was created to radically expand the computational capabilities of AI, complex local agents, and graphically complex games.

NVIDIA RTX Spark processor

The company’s new hardware technology is a highly integrated consumer segment single-chip processor. In fact, RTX Spark is a competitive answer to Qualcomm, AMD and Intel. The architecture of the novelty combines the best server developments of the Grace brand and the custom graphics power of Blackwell on one silicon crystal. Thanks to this solution, developers aim to implement advanced technological scenarios autonomously on the user’s PC without constant access to third-party network infrastructure.

Technical specifications

The company’s modern industrial technology patterns were adapted to create the RTX Spark. Hardware stuffing in its maximum modification surpasses all existing alternatives.

The parameters of the flagship configuration of the novelty look like this:

• Central processing unit (CPU): 20 productive cores of Grace’s own development.
• Graphics (GPU): High performance Blackwell chip with 6144 CUDA computing cores.
• AI Performance: Achieves an impressive 1 PFLOP in machine learning tasks.
• Memory: Integration of up to 128 GB of fast combined memory of the new generation LPDDR5X.

The overall concept of a single data pool facilitates fast addressing. Both the CPU and the graphics chip instantly interact with information from memory without copying data back and forth. Such an innovation guarantees the stability of the system under a heavy mixed load.

Capabilities of local AI and graphics chip

Thanks to the performance reserve of 1 PFLOP, the key vector of the chip’s application is the local processing of resource-intensive scenarios. The system allows fully autonomous and efficient processing of large language models of neural networks (LLM) and the functioning of AI assistants on clients’ computers. In practice, this solution provides an unprecedented level of privacy of the user’s personal metadata, since the system no longer needs external server clouds for successful forecasts.

Implementation of ray tracing and DLSS scaling technologies

Blackwell’s full-fledged integrated graphics unit opens up the launch of demanding video games to thin, energy-efficient devices. The design provides iron-clad compatibility with the standard RTX global light simulation technology (hardware ray acceleration), and the built-in tensor components of the new core easily serve smart smoothing and intelligent DLSS frame rate acceleration technologies on the Windows on Arm architecture.

Release dates

The price has not yet been officially announced by the developer. However, the alliance of companies involved in the development of the project at NVIDIA has formed a really strong one. The biggest players in the computer sector are planning to release equipment in the future. The list of official technology contributors includes the following brands: Acer, ASUS, Dell, Gigabyte, HP, Lenovo, MSI, as well as the software giant Microsoft.

Estimated dates of delivery of the debut commercial equipment of partners with RTX Spark are planned by the developer by the end of this year.