NVIDIA MGX is a new design standard for accelerated computing. It will meet the needs of future servers better than current standards, while reducing the time and cost of deployment and providing maximum flexibility and future-proofing.
MGX utilizes a modular design that permits over 100 different configurations including
GPUs, CPUs, and DPUs, allowing it to meet the needs of nearly any accelerated computing application. In contrast, many current server systems do not provide adequate space for future GPUs, network interface cards, and airflow, or the high TDPs needed for more higher-powered GPUs.
As an MGX chassis partner, Chenbro will provide 19" wide enclosure solutions compliant with the NVIDIA MGX ecosystem standard. Our chassis will allow more flexible PCIe layouts, boast module bays that can support multiple types of server workloads, and provide capacity for new, large GPUs. They can also be adapted for use with future components or for alternative applications, ensuring excellent return on investment. They will also offer features already familiar in Chenbro chassis, such as rear and front cabling options, allowing for server management from either the cold or hot aisle.
In keeping with NVIDIA MGX requirements, configurable modules will be provided in both the rear and front of our chassis. Major subsystems, such as HPM, power distribution boards, and cooling solutions, will all have their own zone within the chassis. Since design parameters will be standardized, modules and components can be switched across applications and generations. The module bays will be available in several sizes while at the same time supporting common form factors and fitting in different-sized chassis.
NVIDIA MGX chassis’ locking and shelving mechanisms will also be standardized so modules can be stacked or interchanged in between chassis. Two locking systems will be used: an internal single spring latch and catch, or an external camming latch.
Chenbro has long been at the forefront of computing technology, having provided chassis for the servers and workstations that form the internet as we know it. We are proud to take part in the next stage of AI computing as an NVIDIA MGX partner.