Home ComputersAINVIDIA RTX Spark Compact Desktops Go on Sale, DGX Spark Kit Links Four Systems

NVIDIA RTX Spark Compact Desktops Go on Sale, DGX Spark Kit Links Four Systems

by Warren
9 views
NVIDIA RTX Spark compact desktop running a local AI creative workflow

NVIDIA is expanding its desktop AI hardware range with RTX Spark compact PCs from ASUS, Dell and MSI now available, while Acer, GIGABYTE, HP and Microsoft systems are due to follow. The company has also introduced a four-node DGX Spark HPC Developer Kit for researchers who want to build and test larger scientific computing and AI workloads at their desks.

The two announcements target different users, but both follow the same idea: keeping more AI work on local hardware instead of sending every prompt, dataset or creative project to a cloud service. NVIDIA has not announced Malaysian pricing or confirmed which configurations will be sold locally, so availability will depend on the respective PC makers and regional channels.

RTX Spark brings up to 128GB of unified memory to compact PCs

RTX Spark desktops combine a 20-core NVIDIA Grace CPU with a Blackwell RTX GPU featuring 6,144 CUDA cores. Configurations can offer up to 128GB of unified memory and deliver up to one petaflop of AI performance, giving the small systems enough headroom to run supported large language models with as many as 120 billion parameters locally.

ASUS ProArt GR1X RTX Spark mini PC in black and white

NVIDIA says the platform can also handle real-time rendering of 3D scenes larger than 90GB and smooth editing of 4:2:2 video. A 10Gb Ethernet connection is included for moving large models, media files and datasets between the PC and network storage.

ASUS is represented by the ProArt GR1X mini PC, while MSI has introduced the EdgeMesa N AI+ series. Dell also has an RTX Spark compact desktop available, with additional systems from Acer, GIGABYTE and HP coming later. Microsoft is preparing a Surface RTX Spark Dev Box for developers.

MSI EdgeMesa N AI Plus RTX Spark compact desktop

For developers, native CUDA support and Windows Subsystem for Linux provide access to familiar Linux tools on a Windows PC. NVIDIA is also positioning the machines as always-on hosts for personal AI agents, with support for Hermes Agent, NVIDIA NemoClaw blueprints and Unsloth fine-tuning workflows. NVIDIA Switchyard can route simpler jobs to local models and send more demanding reasoning tasks to cloud models when needed.

Adobe and Blender add more RTX Spark support

The hardware launch arrives with broader software support. Adobe Premiere uses TensorRT-RTX acceleration for features including Enhance Speech, Object Mask, Scene Edit Detection and Auto Reframe, while Photoshop accelerates its Remove Tool and adds an Arm-native Camera Raw workflow. Illustrator and InDesign now have native Windows on Arm versions, and Substance 3D Painter 12.2 adds native Arm support in public beta.

Blender 5.3 also introduces DLSS 4.5 Ray Reconstruction in the Cycles viewport. This is designed to provide a cleaner, near-final preview of lighting, reflections and materials while an artist works, provided the system is running the required NVIDIA driver.

DGX Spark kit combines four systems for research

At SC26 in Chicago, NVIDIA unveiled the DGX Spark HPC Developer Kit as a four-system cluster. Each DGX Spark node connects directly to two neighbours in a ring through ConnectX-7 networking, removing the need for a separate network switch.

Four NVIDIA DGX Spark systems connected as an HPC developer kit

Together, the four nodes provide 512GB of combined memory and 16TB of NVMe storage. NVIDIA says the cluster can deploy supported models with up to 700 billion parameters using NVFP4, while giving researchers a private platform for developing and debugging scientific workloads before moving them to a shared supercomputer.

Demonstrated workloads include LAMMPS with Kokkos for materials simulation and NAMD for molecular dynamics. The wider DGX Spark community has also tested applications such as GROMACS, OpenFOAM and the WRF weather model. An installation playbook for the NVIDIA HPC SDK is coming later, covering GPU-accelerated C, C++ and Fortran tools as well as multi-node development.

The DGX Spark HPC Developer Kit is available for sale through NVIDIA and its hardware partners, but no Malaysian price has been listed. For local buyers, the key details to watch will be partner availability, support coverage and whether the compact RTX Spark PCs arrive through official ASUS, Dell or MSI channels.

You may also like