Gigabyte G593-ZD2: Maximum GPU Density for AI Workloads
The Gigabyte G593-ZD2 is a 5U GPU server designed to deliver maximum computational density for AI training, deep learning, and high-performance computing. Supporting up to 8 double-width GPUs with full PCIe Gen5 connectivity, this platform provides the raw computing power needed for the most demanding AI and scientific workloads.
Key Specifications
- Processors: Dual AMD EPYC 9004 series (Genoa), up to 192 cores total
- GPU Support: 8x PCIe 5.0 x16 GPU slots (NVIDIA H100/A100, AMD MI300X)
- Memory: 24x DDR5 DIMM slots, up to 6TB at 4800MHz
- Storage: 4x 2.5" NVMe hot-swap + 2x M.2 for OS
- Networking: 2x 10GbE + OCP 3.0 for 100/200GbE InfiniBand or Ethernet
- Cooling: Advanced thermal design with optimized GPU airflow channels
- Power: 4x 3000W redundant PSUs (N+1)
AI Training Performance
With 8 NVIDIA H100 GPUs, the G593-ZD2 delivers over 25 petaFLOPS of AI training performance. The PCIe Gen5 interconnect provides 128GB/s bandwidth per GPU slot, enabling efficient multi-GPU training for large language models, generative AI, and computer vision applications. NVLink bridges can be installed for direct GPU-to-GPU communication at even higher bandwidths.
Deployment Considerations
The 5U form factor accommodates the thermal requirements of 8 high-power GPUs while maintaining datacenter-standard rack compatibility. The advanced cooling system ensures sustained GPU boost clocks under full load, maximizing training throughput and reducing time-to-insight for AI researchers and engineers.
Purchase from Omnixon Global
Omnixon Global offers Gigabyte GPU servers with custom configurations and competitive pricing. Request a quote or explore our Gigabyte server lineup.