They are saying that probably the most troublesome a part of transportation planning is last-mile supply. A community of warehouses and vehicles can deliver merchandise inside a mile of virtually all prospects, however logistical challenges and prices add up rapidly within the means of delivering these items to the appropriate doorways on the proper time.
There’s the same sample within the AI house. Huge information heart installations have empowered astonishing cloud-based AI companies, however many researchers, builders, and information scientists want the ability of an AI supercomputer to journey that final mile. They want machines that provide the comfort and space-saving design of a desktop PC however go effectively above and past the capabilities of consumer-grade {hardware}, particularly in the case of out there GPU reminiscence.
Enter a brand new class of AI desktop supercomputers, powered by ASUS and NVIDIA.
The upcoming ASUS AI supercomputer lineup, spearheaded by the ASUS ExpertCenter Pro ET900N G3 desktop PC and ASUS Ascent GX10 mini-PC, wield the newest NVIDIA Grace Blackwell superchips to ship astounding efficiency in AI workflows. For individuals who want native, non-public supercomputing sources, however for whom a knowledge heart or rack server set up isn’t possible, these programs present a transformative alternative to grab the capabilities of AI.
Scaling up reminiscence to satisfy the parameter depend of huge AI fashions
A key piece of the puzzle for accelerating regionally run AI workloads is out there GPU reminiscence. If a given mannequin doesn’t match into native reminiscence, it could run very slowly, or it could not run in any respect. The 32GB of VRAM supplied by the highest-end NVIDIA consumer-grade graphics card available on the market, the NVIDIA GeForce RTX 5090, is enough for a lot of smaller fashions. However scaling up your system’s VRAM to deal with fashions with much more parameters isn’t essentially an easy affair.
Multi-GPU programs are a possible resolution for some customers, however others have been searching for an answer that’s designed particularly for the wants of AI workflows. By equipping the Ascent GX10 and ExpertCenter Professional ET900N G3 with giant single swimming pools of coherent system reminiscence, we’re in a position to put astonishing portions of reminiscence at your fingertips. The Ascent GX10 wields 4 occasions as a lot GPU reminiscence as a GeForce RTX 5090, whereas the ExpertCenter Professional ET900N G3 affords as much as 784GB — over twice as a lot GPU reminiscence as a workstation geared up with 4 NVIDIA RTX PRO™ 6000 GPUs.
AI supercomputer efficiency in a desktop PC type issue
Designed from the bottom up for AI workflows, the ASUS ExpertCenter Pro ET900N G3 shall be one of many first pioneers in a brand new class of computer systems, primarily based on the NVIDIA DGX station.
This method is powered by the NVIDIA GB300 Grace Blackwell Extremely Desktop Superchip. That includes an NVIDIA Blackwell Extremely GPU and an NVIDIA Grace CPU related through the NVIDIA® NVLink®-C2C interconnect, this superchip supplies a slice of information heart efficiency in a desktop workstation. Much more so than at present’s high-end desktop programs, the ExpertCenter Professional ET900N G3 ensures that companies and researchers can develop and run large-scale AI coaching and inference workloads because of as much as 784GB of huge coherent reminiscence.
All of it runs on the NVIDIA AI Software program Stack together with NVIDIA DGX OS, a personalized set up of Ubuntu Linux purpose-built for optimized efficiency in AI, machine studying, and analytics purposes, with the flexibility to simply scale throughout a number of NVIDIA DGX Station programs.
The AI supercomputer within the palm of your hand: the ASUS Ascent GX10
The ASUS ExpertCenter Professional ET900N G3 is far simpler to deploy than an answer primarily based on rack servers, however there are conditions the place even a desktop-class type issue remains to be too giant. The ASUS Ascent GX10 democratizes AI by placing petaflop-scale AI computing capabilities in a design which you could maintain within the palm of your hand.
The ASUS NUC lineup demonstrates our confirmed experience in providing full PC experiences in ultracompact designs. No mere iterative step ahead, the Ascent GX10 takes our expertise within the mini-PC market and melds it with the groundbreaking efficiency of the NVIDIA GB10 Grace Blackwell Superchip. This superchip connects a Grace CPU with 20 Arm cores with a sturdy Blackwell GPU by way of NVIDIA® NVLink®-C2C know-how. All advised, it delivers as much as 1,000 AI TOPS of processing energy, with 128GB of coherent unified system reminiscence permitting the system to deal with 200 billion parameter AI fashions.
Want the Ascent GX10 to deal with even bigger fashions, comparable to Llama 3.1 with its 405 billion parameters? Built-in NVIDIA® ConnectX®-7 Community Expertise permits you to harness the AI efficiency of two Ascent GX10 programs working collectively.
A part of an entire AI resolution set
ASUS stands out from each different producer available on the market with the breadth of AI merchandise that we’re in a position to supply. The ExpertCenter Professional ET900N G3 and Ascent GX10 slot into an entire lineup that meets the wants of AI fans at each degree. For these trying to build their own AI PC out of consumer-grade parts, for individuals who want AI performance built into their everyday laptop, for enterprises who want a single-rack AI server solution, even for these establishments trying to design, deploy, and operate a data center for AI applications, the ASUS product portfolio is prepared.
But the ExpertCenter Professional ET900N G3 and Ascent GX10 are way over mere additions to our AI product stack. The soar from AI PC to AI supercomputer is nothing lower than revolutionary, and these programs provide you with this degree of efficiency in an entire turnkey resolution that matches on a desktop.
Features of those programs are nonetheless in improvement, however we’ll share extra particulars as quickly as we’re ready. Within the meantime, explore how ASUS can help your organization seize the capabilities of AI.
