AI and HPC Simplified
GigaIO SuperNODE™
The World’s First 32 GPU Single-node Server
“Less time messing with infrastructure — faster time to running LLMs and optimizing.”
— Greg Diamos, Co-founder & CTO, Lamini
Meet the Creator of FluidX3D and See One of the Largest
CFD Simulations
SuperNODE, the world’s first 32 GPU server, flawlessly executed one of the largest CFD simulations ever — the Concorde landing at 40 BILLION cells resolution — in a mere 33 hours. The secret? Massive GPU memory, making the complex fast and easy.
Daily Demo:
See the SuperNODE Shatter the 8-GPU Barrier!
Stop by the AMD/GigaIO booth (#F19) for a live demo showing the SuperNODE with AMD MI300X GPUs running LLMs on a single node.
Monday: 18:00
Tuesday: 14:00
Wednesday: 14:00
Can’t make one of these times? Check out the AMD/GigaIO On Demand Webinar
SuperNODE In the News
See For Yourself
Join GigaIO at an ISC BoF:
Increasing Memory Utilization and Reducing Total Memory Cost Using CXL
Come discuss the release of the CXL 3.1 specs and explore CXL features, use case enablement, and ROI examples when implementing CXL attached memory.
Monday, May 13 — 17:40-18:40 in Hall F (second floor)
Senior Director of Engineering, GigaIO
GigaIO SuperNODE: The World’s First 32 GPU Single-node Supercomputer
The GigaIO SuperNODE can connect up to 32 AMD or NVIDIA GPUs to a single node at the same latency and performance as if they were physically located inside the server box. Accelerators are seamlessly connected by GigaIO’s transformative PCIe memory fabric, FabreX, and can scale multiple accelerator technologies such as GPUs and FPGAs without the latency, cost, and power overhead required for multi-CPU systems.
See demo at our booth.
Revolutionize CFD with the SuperNODE
The world’s first 32 GPU server flawlessly executed one of the largest CFD simulations ever — the Concorde landing at 40 BILLION cells resolution — in a mere 33 hours. The secret? Massive GPU memory, making the complex fast and easy. Learn how the SuperNODE has revolutionized CFD with an innovative approach to solving for extreme-scale CFD in this remarkable case study. Stop by our booth to meet the FluidX3D developer, who performed this incredible benchmark.
Meet the FluidX3D developer at our booth.
The GigaIO Family Of Engineered Solutions, Powered By AMD
Make Science Simple. GigaIO has partnered with AMD to release a family of purpose-built Engineered Solutions that use composability to solve industry-specific challenges for Life Sciences, Manufacturing, and Education. Get the right hardware and software for each vertical — easy to install and easy to administer with NVIDIA BCM pre-installed.
Learn more at our booth.
FabreX Universal Dynamic Memory Fabric
Let your software define your hardware. GigaIO provides workload-defined infrastructure through its universal dynamic memory fabric, which seamlessly composes rack-scale resources and integrates natively into industry-standard tools. Build impossible servers for HPC + AI workflows — from storage to accelerators to memory — at a fraction of cloud TCO, by optimizing the utilization and efficiency of your existing hardware.
See demo at our booth.
