- Asus ExpertCenter Pro Et900N G3 hides the wildest nvidia chip
- With up to 784 GB of memory, it manages models that its RTX 5090 simply cannot
- Without rack, without noise, only a supercomputer class performance on a desktop you can use
ASUS has presented a new high -performance desktop PC that offers a Petaflop scale yield, not on a striking server shelf, but in a surprisingly unpretentious chassis.
The ASUS Expertcenter Pro ET900N G3 resembles any standard business tower, so much that it even includes a DVD unit and a curious appearance groove that feels like a setback in the early 2000s.
In the heart of this commercial PC is the NVIDIA GB300 Ultra, a two-part module that combines a Grace CPU with a Blackwell GPU through NVLink-C2C, the high bandwidth interconnection of NVIDIA, which makes it ideal for the development of programming and AI.
Designed for serious work of AI
The unified chips architecture allows the CPU and the GPU to share a single memory group, reducing latency and improving efficiency for large -scale AI workloads. The system can deliver up to 20 performance PFLOP to train large language models or execute inference in high parameter models.
It admits up to 784 GB of coherent memory, more than double the combined vram of a workstation with four ada RTX 6000 cards.
This access scale is essential for developers and researchers who work with models that exceed conventional GPU capabilities such as the GeForce RTX 5090, which offers 32 GB of VRM.
Expertcentr Pro ET900N G3 also includes support for NVIDIA Connectx-8 Supernic, which allows the high performance network among the systems. This allows you to function in groups or within a larger business implementation.
Despite its performance, it retains a desktop form, eliminating the need to install frame, custom cooling solutions or the infrastructure demands of a data center.
On the software side, the system executes NVIDIA DGX OS, a specialized linux distribution based on Ubuntu designed for AI work loads. It provides native support for the full NVIDIA software battery, including CUDA, Tensorrt and libraries for automatic learning and data science.
It is also compatible with the remote scale, allowing the ET900N G3 to integrate perfectly with other DGX systems if additional computing energy is needed.
