May 12, 2024 – In a joint effort, technology firms GigaIO and SourceCode have unveiled a revolutionary portable AI supercomputer dubbed Gryf. Gryf stands out due to its compact design, resembling the size of a carry-on suitcase and weighing no more than 55 pounds, adhering to the TSA guidelines for carry-on luggage, thereby offering unmatched portability.
Despite its diminutive size, Gryf packs a punch with four graphics processors (GPUs), a whopping 246TB of storage space, and an integrated power supply capable of delivering up to 2500 watts. This robust configuration meets the demanding data collection and processing needs of AI projects, traditionally reliant on large data centers, which often entailed time-consuming data transfers and potential security risks. Gryf effectively addresses these challenges.
One of Gryf’s most innovative features is its customizable hardware configuration. Leveraging GigaIO’s FabreX AI memory architecture, users can swap internal modules based on specific workloads. For AI or machine learning tasks, users can opt for a setup with two computing modules, an accelerator module, two storage modules, and a networking module. Alternatively, for storage-intensive projects, the configuration can be adjusted to include one computing module and five storage modules.
According to GigaIO and SourceCode, a single Gryf unit can handle over a petabyte of information. Moreover, utilizing the FabreX memory architecture allows up to five Gryf units to be stacked, enabling collaborative processing for even more complex tasks. Once on-site data collection and preliminary processing are complete, Gryf can be easily transported back to the data center and connected to GigaIO’s more powerful core computer, GigaPod, for deeper analysis.
GigaIO CEO Alan Benjamin has emphasized Gryf’s potential in defense applications. Its portability and performance make it ideal for military personnel who require real-time, actionable intelligence in field environments. Gryf’s customizable hardware and FabreX-supported innovative architecture allow it to adapt to various sensor-intensive edge computing scenarios, providing users with cutting-edge computing, storage, and graphics processing capabilities.
Currently, Gryf is available for purchase, although the specific pricing details remain undisclosed.
