Gryf co-designed by SourceCode and GigaIO is a portable supercomputer for rapid AI processing of time-sensitive data in the field. Gryf redefines portability by packing a powerful datacenter into a ruggedized suitcase form factor.
Gryf is a singular, composable solution for a diverse set of mission needs. Fully configurable on-demand through software or by interchanging sleds, Gryf allows for dynamic reconfiguration, resulting in higher resource utilization, reduced hardware costs, higher availability, and lower power consumption.
Other highlights of the unique capabilities Gryf offers include:
The Gryf system is built to be highly customizable at purchase. And reconfigurable in the field.
It is designed from the ground up to fit your unique workload, budget, and deployment environment.
Below are three common configuration profiles tailored for AI/ML/LLM, high-performance computing, and storage-intensive operations. Each showcases a different balance of compute, acceleration, storage, and networking, demonstrating how Gryf adapts to specific mission needs.
These configurations are ready to deploy or can serve as flexible starting points, easily modified to fit your needs. Wherever you operate, our team can help fine-tune the right Gryf for your application.
Sample configuration built for inference, training, and edge deployment of AI workloads in the field.
(1) Enclosure I – Carbon Fiber case with removable top and bottom covers
(1) Network Sled I – 2 x 100GbE QSFP56 + 6 x 25/10GbE SFP28 Ports
(1) NVIDIA L40S GPU
(1) Storage Sled I – 8 x 30TB NVMe Drives
(1) Compute Sled I – 16 CORE + 256GB DDR4
Optional Extended Warranty available
Optional Installation Services available
Sample configuration built for simulation, modeling, and parallel processing at the tactical edge.
(1) Enclosure I – Carbon Fiber case with removable top and bottom covers
(1) Network Sled I – 2 x 100GbE QSFP56 + 6 x 25/10GbE SFP28 Ports
(1) Storage Sled I – 8 x 30TB NVMe Drives
(2) Compute Sled I – 16 CORE + 256GB DDR4
Optional Extended Warranty available
Optional Installation Services available
Sample configuration built for large-scale data capture, local storage, and high-throughput transfer.
(1) Enclosure I – Carbon Fiber case with removable top and bottom covers
(2) Storage Sled I – 8 x 30TB NVMe Drives
(1) Compute Sled I – 16 CORE + 256GB DDR4
Optional Extended Warranty available
Optional Installation Services available
Connect up to five Gryf units (with 30 available slots for resources) using GigaIO’s AI Fabric Manager to achieve supercomputer-class performance. AI fabric technology dynamically disaggregates and re-aggregates GPUs and other system components to optimize hardware configurations for each workload. This setup allows each Gryf unit to share and access sled components across all five interconnected chassis in an all-to-all network configuration.
Our co-design approach revolutionizes solution creation by bringing the customer together with technology experts and best-of-breed products to develop and deploy the perfect solution. The Gryf co-design partnership blends GigaIO’s extensive knowledge in interconnects and composable infrastructure with SourceCode’s expertise in ruggedized system design, co-design manufacturing, and global SourceCode Labs (technology testbed centers).
Elevate your edge computing capabilities with Gryf, the ultimate in portable solutions for mission-critical operations.
Discover a range of third-party integrations that elevate the functionality and flexibility of Gryf. Our marketplace partners provide innovative solutions that seamlessly work with Gryf, allowing you to optimize your workflows and maximize performance.
Gryf, powered by Ipsotek VISuite, is a fully portable, edge-based AI video analytics solution that delivers real-time intelligence for security, operations, and surveillance—anywhere it’s needed, without relying on the cloud.
The Hammerspace Global Data Platform can be used to process, store, and orchestrate data in any edge computing environment, including those operating with limited or no networking infrastructure.
The VAST Data Platform is a comprehensive software infrastructure for capturing, cataloging, refining, enriching, and preserving data, while enabling real-time data analysis and deep learning on unstructured data from anywhere.
WEKA offers unparalleled speed and flexibility for demanding applications such as AI, GPU acceleration, HPC. It can handle tens of millions or even billions of files of all data types and data sizes and can be deployed at the edge.
