Gigabyte SV G492-HA0 4U DP 10xGPU 2xLGA4189 Socket P+ C621A DDR4 3x2200W RTL

Where Compute Reigns | 3rd Gen Intel® Xeon® Scalable Processors - "Ice Lake"

GIGABYTE servers and Intel Xeon Scalable Processors deliver incredible 1P/2P performance from edge to data center with incredible gains in I/O throughput and workload demands that need high performing CPUs with large, yet optimized, memory configurations.

• Enhanced I/0: Support for PCIe 4.0 allows for double the throughput of PCIe 3.0 for fast and large data transfers. Also, 64 lanes per socket with up to 128 lanes in a 2P configuration for fast data transmission from CPU to GPU, accelerator or storage.
• CPU Performance Boost: Now up to 40 cores per socket with greatly improved IPC on Intel's 10nm architecture while operating CPUs at 105-270W. Also, an additional UPI lane is added for up to 11.2GT/s for a low latency interconnect between CPUs.
• Next Gen Memory: Support for DDR4-3200MHz (1DPC or 2DPC) DIMM. 8 channel memory has been added with up to 16 DIMMs per socket delivering up to 4TB of system memory via usage of 256GB DDR4 DIMM. Additionally, 2/4/6/8-way interleaves for optimal configuration.
• Arrival of Intel Optane PMem 200 Series: Optimized for improvements in latency while increasing system memory up to 6TB per socket as a mix of PMem and DDR. The 200 Series also has support for memory mode to expand capacity and app direct mode to act as fast flash storage.
• Next Level Security: New instructions and architecture deliver a high level of cryptographic operations for data privacy and protection. Cryptographic accelerators speed up encryption protocols as well as improvements in software guard extensions and memory encryption are added.
• AI Acceleration: AI inference and training are enhanced with DL Boost Technology for greater business, operational, and security insights. Applications include HPC, AI, and Media & Graphics.

Why GIGABYTE G492-HA0 | Accelerator Computing | Dual root complex

Parallel processing with multiple GPUs and support for dual slot GPUs allows for the highest, dense GPU compute possible. To get even more GPU performance a balanced dual root complex allows for offers the lowest latency as system memory is accessed by GPU memory via PCIe switch, while bypassing the CPU.

High-performance | NVLink and GPUDirect utilized

Requiring the ultimate performance in every aspect of a server is possible. GIGABYTE servers have been designed to support the cutting edge, high-end CPUs, GPUs P2P, and GPU-to-NIC bridges while utilizing GPUDirect and PCIe 4.0 for fast throughput and multi-GPU communication.