Menu
QCT Server Solutions Powered by AMD Instinct MI325X accelerator
QCT Server Solutions, powered by the latest AMD Instinct™ MI325X accelerators, delivers cutting-edge performance with 288GB of high bandwidth memory (HBM3E) and an impressive memory bandwidth of 6 terabytes per second. It utilizes the industry-standard-based universal baseboard (UBB 2.0) server design that is consistent with the AMD Instinct™ MI300 series.
Looking ahead, the next-generation AMD Instinct™ MI350 Series CPUs and platforms, built on the AMD CDNA™ 4 architecture, are set for release in 2025. Featuring advanced 3nm process technology, they will support FP4 and FP6 AI data types while offering up to 288GB of HBM3E memory. They will also remain compatible with the universal baseboard server design to ensure seamless integration and scalability.
QuantaGrid D75T-7U, equipped with AMD EPYC™ 9004/9005 Series processors and AMD Instinct™ MI325X GPUs, is engineered for unparalleled AI training and inference efficiency. Its future-ready design, supporting upcoming AMD Instinct™ MI350 GPUs and platforms, enables sustainable scaling of AI deployments in next-gen data centers.
QuantaGrid D75M-5U is a liquid cooled solution powered by the latest AMD EPYC™ 9004/9005 Series processors. It utilizes liquid cooling to achieve optimized power consumption for different AI workloads. This 5U system is designed to excel in demanding AI tasks like training expansive models and inference, delivering incredible performance and efficiency for Gen AI models.
QuantaGrid D74A-7U is powered by the latest AMD EPYC™ 9004 Series processors and AMD Instinct™ MI325X GPU accelerators to deliver optimized performance for AI training. Offering drop-in compatibility with existing AMD Instinct™ MI300X infrastructures, QCT’s QuantaGrid D74A-7U boosts performance for AI inference and fine-tuning without requiring infrastructure changes. As one of the most powerful accelerated server platforms available, it is built for massive AI models, including generative AI and large language models (LLMs), as well as advanced scientific computing for HPC workloads.
QuantaGrid S44NL-1U
QuantaGrid D44N-1U
QCT Next Gen General Purpose Servers
QCT general purpose servers powered by the latest AMD EPYC™ Processors offer ultimate compute and workload-optimized performance. Featuring QCT’s forward-thinking designs and robust performance capabilities, these servers help our customers to confidently scale ahead to tackle the evolving demands of tomorrow’s AI world.
AMD Instinct™ MI325X Accelerators
Model Specification
- Memory: Up to 256GB of HBM3E memory
- Memory bandwidth: 6TB/s peak memory bandwidth
- Architecture: AMD CDNA™ 3
- Chiplets: Eight compute, four I/O, and eight memory chiplets using advanced 2.5D and 3D packaging technology
Product Features
- Built for enterprise AI, offering seamless integration for fast, cost-effective upgrades
- Drop-in compatible with existing AMD Instinct™ MI300X infrastructures
- Infinity Fabric™ technology for seamless data sharing
- AMD ROCm™ software to accelerate AI inferencing and training
AMD Technology Features
AMD EPYC™ 9005 Series
AMD EPYC™ 9004 Series
- Up to 128 Zen5 cores / 256 threads; 192 Zen5c cores / 384 threads
- Largest available L3 cache up to 512 MB
- TDP up to 500 W
- Supports up to 32 GT/s xGMI
- 12-channel DDR5 RDIMM/ 3DS RDIMM, 2 DIMM per channel
- Memory frequency of up to 6400 MHz
- Supporting 16Gb, 24Gb, 32Gb devices
- 128 lanes PCIe 5.0 of 32 Gb/s bandwidth
- Up to 4 xGMI-3 coherent interconnect with 32 GT/s bandwidth
Elevated Data Center Performance with AMD EPYC ™ Series Processors
QCT Product Info
SDS/HCI
Virtualization
Design &
Simulation
IC Design &
EDA
Machine
Learning
Cloud Gaming/
Rendering
AMD, the AMD arrow logo, EPYC, and combinations thereof are trademarks of Advanced Micro Devices, Inc.
#Compute
#AI Inference
#GenAI
#HPC
#EDA
#CFD
#Storage
QuantaGrid D74A-7U
Advanced Performance for the Most Extreme AI-HPC Workloads
- Powered by dual AMD EPYC™ 9004 Processors with up to 500W TDP and AMD Instinct™ MI300X GPU accelerators.
- UBB architecture to support AI & HPC workloads
- PCIe 5.0 OCP expansion slots for RDMA networking
- Modularized system design for optimal flexibility and easy serviceability
- High NVMe storage density, supporting up to 18x 2.5” NVMe drives.
- Accelerated parallel computing performance for the most extreme AI-HPC workloads
- Enhanced serviceability with tool-less, hot-pluggable designs
| Processor | (2) AMD EPYC™ 9004 Processors, up to 400W TDP | Memory | Up to 6 TB memory capacity with (24) DDR5 RDIMM up to 4800 MHz | Networking | Dedicated (1) GbE management port | Storage | Front (18) 2.5″ hot-pluggable NVMe drives | Expansion Slot | (2) PCIe 5.0 x16 OCP 3.0 SFF slots (10) PCIe 5.0 x16 OCP 3.0 TSFF slots |
GPU Expansion Slot | (8) Accelerator modules on one single baseboard (AMD Instinct™ MI300X and NVIDIA HGX™ H100 8-GPU baseboard) |
Dimensions | 447.8 mm x 307.85 mm x 886 mm | Form Factor | 7U rackmount server |
#Compute
#AI Inference
#GenAI
#HPC
#EDA
#CFD
#Storage
QuantaGrid D75T-7U
Transform Computational Capabilities and Achieve High Performance
- Support AMD Instinct™ MI350 Series GPUs and platforms
- Powered by dual AMD EPYC™ 9005 Series Processors with up to 500W TDP and AMD Instinct™ MI300X/MI325X GPU accelerators
- UBB architecture to support AI & HPC workloads
- PCIe 5.0 expansion slots for RDMA networking
- Modularized system design for optimal flexibility and easy serviceability
- High NVMe storage density, supporting up to 18x 2.5” NVMe drives.
- Accelerated parallel computing performance for the most extreme AI-HPC workloads
- Enhanced serviceability with tool-less, hot-pluggable designs
| Processor | (2) AMD EPYC™ 9005 Series Processors, up to 500W TDP | Memory | Up to 6 TB memory capacity with (24) DDR5 RDIMM up to 6400 MHz | Networking | Dedicated (1) GbE management port | Storage | Front (18) 2.5″ hot-pluggable NVMe drives | Expansion Slot | (2) PCIe 5.0 x16 FHHL slots (8) PCIe 5.0 x16 HHHL slots |
GPU Expansion Slot | (8) Accelerator modules on one single baseboard (AMD Instinct™ MI300X/ MI325X 8-GPU baseboard) |
Dimensions | 447.8 mm x 307.85 mm x 950 mm | Form Factor | 7U rackmount server |
#Compute
#AI Inference
#GenAI
#HPC
#EDA
#CFD
#Storage
QuantaGrid D75M-5U
Advanced Performance for the Most Extreme AI-HPC Workloads
- Powered by AMD EPYC™ 9004 or 9005 Processors with up to 500W TDP
- Liquid cooling solution designed for AMD CPU platforms
- Accelerated parallel computing performance for the most extreme AI-HPC workloads
| Processor | (2) AMD EPYC™ 9004/9005 Series Processors, up to 500W TDP | Memory | Up to 6 TB memory capacity with (24) DDR5 RDIMM up to 6400 MHz | Networking | Dedicated (1) GbE management port | Storage | Front (18) 2.5″ hot-pluggable NVMe drives | Expansion Slot | (2) PCIe 5.0 x16 FHHL slots (8) PCIe 5.0 x16 HHHL slots |
GPU Expansion Slot | (8) Accelerator modules on one single baseboard (NVIDIA HGX™ H200 8-GPU baseboard) |
Dimensions | 447.8 mm x 219.5 mm x 950 mm | Form Factor | 5U rackmount server |
#Compute
#AI Inference
#HPC
#EDA
#CFD
#Storage
QuantaGrid S44NL-1U
Cost-effective Server with Scalable Expansion Options and Optimize Performance
- Supports 1x AMD EPYC™ 9004/9005 Processor, up to 500W TDP
- Advanced air-cooling architecture supporting top-bin CPUs
- Liquid cooling-ready by design
- Up to 5x PCIe 5.0 expansion slots and the DC-SCM architecture to meet different configuration needs
- Featuring up to 2x single-width accelerators to support AI inference workloads
- System optimized for cloud, enterprise, AI, HPC, networking, security and IoT workloads
| Processor | (1) AMD EPYC™ 9004/9005 Series Processor, up to 500W TDP | Memory | Up to 6 TB memory capacity with (24) DDR5 RDIMM up to 5200 MHz | Networking | Dedicated (1) GbE management port | Storage | All U.2 Front (12) 2.5″ hot-pluggable NVMe/SATA/SAS drives All E1.S Front (12) or (16) E1.S hot-pluggable NVMe drives |
Expansion Slot |
Option 1-1 (2) PCIe 5.0 x16 FHHL slots (2) PCIe 5.0 x16 OCP 3.0 slots Option 1-2, 2-1 (2) PCIe 5.0 x16 HHHL slots (2) PCIe 5.0 x16 OCP 3.0 slots Option 1-3. 2-2 (3) PCIe 5.0 x16 HHHL slots (2) PCIe 5.0 x16 OCP 3.0 slots |
Dimensions | 440 mm x 43.2 mm x 780 mm | Form Factor | Rackmount server |
#Compute
#AI Inference
#HPC
#EDA
#CFD
#Storage
QuantaGrid D44N-1U
Ultimate 1U Server with AMD EPYC Performance Breakthrough
- Powered by dual AMD EPYC™ 9004/9005 Series Processors, up to 500W TDP
- Advanced air-cooling architecture supporting top-bin CPUs
- Liquid cooling-ready by design
- Up to 5x PCIe 5.0 expansion slots and the DC-SCM architecture to meet different configuration needs
- Featuring up to 2x single-width accelerators to support AI inference workloads
- System optimized for cloud, enterprise, AI, HPC, networking, security and IoT workloads
| Processor | (2) AMD EPYC™ 9004/9005 Series processors, up to 500W TDP | Memory | Up to 6 TB memory capacity with (24) DDR5 RDIMM up to 6400 MHz | Networking | Dedicated (1) GbE management port | Storage |
All U.2 Front (12) 2.5″ hot-pluggable NVMe/SATA/SAS drives All E1.S Front (12) or (16) E1.S hot-pluggable NVMe drives |
Expansion Slot |
Option 1-1 (2) PCIe 5.0 x16 FHHL slots (2) PCIe 5.0 x16 OCP 3.0 slots Option 1-2, 2-1 (2) PCIe 5.0 x16 HHHL slots (2) PCIe 5.0 x16 OCP 3.0 slots Option 1-3. 2-2 (3) PCIe 5.0 x16 HHHL slots (2) PCIe 5.0 x16 OCP 3.0 slots |
Dimensions | 440 mm x 43.2 mm x 780 mm | Form Factor | Rackmount server |
Download Brochure
QCT Next Gen Server Solutions Powered by AMD EPYC™ 9005 Series Processors (EN)
Select your languages: