The Gigabyte G492-ZD2 GPU server is a 4U dual-socket AMD EPYC processor server with support for the latest technology. There are a total of 6x hot-swappable 2.5-inch Gen4 U.2 NVMe or SATA bays on the top half of the chassis, while a pull-out drawer underneath the motherboard houses up to 8x NVIDIA A100 Tensor Core GPUs.
Supporting AMD EPYC 7002 or 7003 series processors, the server features 32x memory module slots with a total capacity of up to 8TB of memory, depending on the type of DIMMs used. The GPU server is equipped with 6x drive bays on the front panel, plus 2x M.2 slots on the motherboard for additional storage or to support the OS. This system is ideal for artificial intelligence, AI training and inference, plus high-performance computing. The system is powered by 4x 80 PLUS Platinum 3000W PSUs operating in a 3+1 redundancy.
Gigabyte G492-H80 GPU Server
2x 3rd Gen Intel Xeon Scalable, 8x PCIe 4.0 x16, 12x SATA
The Gigabyte G492-ZD2 GPU server is a 4U NVMe-enabled server that is built around AMD EPYC 7002 or 7003 series processors to deliver the highest possible I/O throughput. The top half of the server holds the motherboard with dual CPUs each supporting up to 64 cores and 128 threads. Each socket can support a thermal design point (TDP) of up to 240W and a maximum cTDP of up to 280W. Up to 32x DIMM slots are divided equally between the dual processors with memory speeds of up to 3200MHz supported. The system is enabled with automatic fan speed control and will be adjusted according to temperature sensors.
The motherboard on the Gigabyte G492-ZD2 (Rev. A00) supports 8x memory channels and provides 16x DIMMs per socket for a total of 32x DDR4 memory slots. When all slots are populated, RDIMM and LRDIMM modules will support up to 4TB of memory. 3DS RDIMM and 3DS LRDIMM modules will support up to 8TB of memory at full capacity. When using 2x DIMMs per channel (DPC), the server can still support memory speeds of up to 3200MHz.
All 6x 2.5” Gen4 drive bays on the front of the system are hybrid and can be wired for SATA or U.2 NVMe. These storage bays are hot-swappable and can be configured with hard disk drives or SDDs.
On the bottom half of the server there is a pull-out drawer with NVIDIA’s HGX A100 supercomputing delta baseboard to tackle massive datasets and complex simulations. The platform features 8x SXM4 NVIDIA A100 Tensor Core 80GB modules, providing up to 1.3TB of GPU memory and over 2TB/s of memory bandwidth for unrivaled acceleration. Nvidia’s NVLink and NVSwitch technology is supported, and supports a GPU-to-GPU interconnect of up to 600Gb/s. On the rear of the chassis, there are 8x low-profile Gen4 x16 slots, plus 2x low-profile PCIe Gen4 x16 slots on the front panel, for a total of 10x PCIe expansion slots. In addition, there are 2x ultra-fast M.2 slots with PCIe Gen4 x4 interface supporting 2242, 2260, and 2280 cards.
The Gigabyte G492-ZD2 GPU server allows researchers to deliver real-world results quickly and deploy solutions at scale. Because of this, the server offers protection from security attacks and cyber threats by providing a secure application environment during runtime. Server health monitoring, remote trouble shooting, and a reduction in server downtime is included free-of-charge via the baseboard management controller (BMC) built into the server plus Gigabyte’s proprietary Management Console.
The Gigabyte G492-ZD2 (Rev. A00) GPU server is a high-performance computing platform for data centers. This AMD EPYC-based system supports second or third generation processors, plus PCIe Gen 4 technology and NVMe drive connectivity. Adding the NVIDIA HGX platform brings together the full power of NVIDIA GPUs to accelerate emerging workloads and tackle massive datasets.
-M-Key
-PCIe Gen4 x4
-Supports NGFF-2242/2260/2280 cards
If you know what you want but can't find the exact configuration you're looking for, have one of our knowledgeable sales staff contact you. Give us a list of the components you would like to incorporate into the system, and the quantities, if more than one. We will get back to you immediately with an official quote.