Quanta Cloud Technology (QCT) has announced a new lineup of QuantaGrid servers powered by the upcoming AMD EPYC 9006 Series CPUs. These systems target the increasing demands of agentic AI, high-performance computing (HPC), and modern data center workloads. By leveraging the 6th Gen AMD EPYC processors, QCT aims to provide organizations with the dense compute performance required to scale AI agents and maximize infrastructure efficiency.
QuantaGrid D76W-2U and S46V-1U Specifications
The new QuantaGrid platforms include the S46V-1U and the D76W-2U. The S46V-1U is a 1U server designed for enterprise and cloud environments, featuring a single AMD EPYC 9006 CPU, 16 DDR5 memory channels, and support for up to four PCIe Gen6 expansion slots. It accommodates various storage configurations, including U.2, E1.S, and E3.S SSDs, with a tool-less, hot-pluggable design for serviceability.
The D76W-2U is a 2U liquid-cooled busbar system utilizing dual AMD EPYC 9006 CPUs paired with NVIDIA HGX Rubin NVL8. This configuration integrates NVIDIA BlueField-4 DPUs to offload networking, storage, and security services, alongside NVIDIA ConnectX-9 SuperNICs for high-bandwidth, low-latency connectivity. These systems utilize 2nm process technology and the "Zen 6" architecture to deliver up to 256 cores and 1.6 TB/s of memory bandwidth.
Scaling Agentic AI and HPC Workloads
These servers are engineered to address the shift toward smaller, distributed AI models. By utilizing the 6th Gen AMD EPYC CPUs, QCT provides higher VM density and increased AI service throughput per rack. This architecture is intended to accelerate token generation and improve throughput for a wide spectrum of general-purpose and AI-first workloads.
Key Takeaways
- The new QuantaGrid platforms utilize AMD EPYC 9006 Series CPUs built on the "Zen 6" architecture.
- The QuantaGrid D76W-2U features liquid cooling and integrates NVIDIA HGX Rubin NVL8 and BlueField-4 DPUs.
- The systems support PCIe Gen6 and provide memory bandwidth of up to 1.6 TB/s.
TechInsyte's Take
In our view, QCT’s focus on the 6th Gen AMD EPYC architecture signals a strategic pivot toward the specific hardware requirements of agentic AI. By integrating NVIDIA’s HGX Rubin and BlueField-4 DPUs into the D76W-2U, QCT is positioning its infrastructure to handle the complex orchestration and high-bandwidth connectivity required for distributed AI models. This approach suggests that as AI workloads evolve, the industry will prioritize specialized, liquid-cooled, and high-density configurations to manage the intense computational and networking demands of next-generation AI factories.
Source: https://www.einpresswire.com/
