High‑Density Dual‑Socket AMD EPYC Server — Built for Extreme Throughput and Scalability

Unleash top‑tier performance for AI/ML, virtualization, databases, analytics, and high‑density storage with this dual‑socket AMD EPYC platform. Engineered on 7nm technology, it delivers massive core counts, PCIe Gen4 bandwidth, and enterprise‑class redundancy—ready for 24/7 data center workloads in Dhaka and beyond.

Compute Powerhouse

• Dual AMD EPYC 7003/7002 Series processors with up to 64 cores and 128 threads per CPU for up to 128 cores/256 threads per node.

• 7nm architecture for excellent performance‑per‑watt and dense compute efficiency.

• TDP up to 225W (cTDP up to 240W), with support for CPU 280W at 30°C ambient for demanding workloads.

• Optimized for multi‑GPU deployments via PCIe Gen4 x16 lanes.

Massive Memory Capacity & Bandwidth

• 32 DIMM slots with 8‑channel memory architecture for extreme parallel throughput.

• DDR4 only; speeds up to 3200/2933MT/s, ideal for memory‑bound applications.

• Flexible module support:

  • RDIMM up to 128GB

  • LRDIMM up to 128GB

  • 3DS RDIMM/LRDIMM up to 256GB

• Scale to multi‑terabyte memory footprints to accelerate in‑memory databases and virtualization density.

Flexible, Hot‑Swap Storage

• 12x front hot‑swappable bays for 3.5″/2.5″ SATA/SAS drives.

  • 8 bays connected to CPU_0, 4 bays to CPU_1 for balanced I/O paths.

• Default options:

  • 12x SATA drives, or

  • Hybrid: 4x SATA + 8x SAS via SAS HBA

• Ideal for tiered storage, software‑defined storage, and high‑IOPS arrays.

High‑Speed Networking & Management

• Rear I/O:

  • 2x 1GbE LAN (Intel I350‑AM2) with NCSI support for streamlined out‑of‑band control.

  • 1x dedicated 10/100/1000Mbps management LAN for always‑on IPMI/BMC access.

• Expansion options:

  • 1x OCP 3.0 slot (PCIe Gen4 x16) from CPU_0 with NCSI support for flexible NIC upgrades.

  • 1x OCP 2.0 mezzanine (PCIe Gen3 x8, Type1 P1/P2) from CPU_1 with NCSI support.

GPU‑Ready PCIe Gen4 Expansion

• Riser CRS2014:

  • 1x PCIe Gen4 x16 FHFL from CPU_1 (GPU‑optimized).

• Riser CRS2026 x2:

  • 1x PCIe Gen4 x16 FHFL from CPU_0 (GPU‑optimized)

  • 1x PCIe Gen4 x16 FHHL from CPU_0

  • 1x PCIe Gen4 x16 FHFL from CPU_1 (GPU‑optimized)

  • 1x PCIe Gen4 x16 FHHL from CPU_1

• Design supports multi‑GPU configurations, NVMe adapters, high‑speed NICs, and storage HBAs.

M.2 Boot/Cache Flexibility

• 1x M.2 (M‑key) PCIe Gen3 x4 from CPU_0.

• Supports 2242/2260/2280/22110 form factors for OS boot or cache tiers.

• Note: Using M.2 limits CPU TDP to 225W for thermal balance.

Data Center‑Grade Power and Redundancy

• Dual 2000W hot‑swap redundant PSUs with high efficiency:

  • 80 PLUS Platinum (-A00) option with wide AC/DC input support and tiered DC output profiles for 1000W/1600W/1800W/2000W ranges.

  • 80 PLUS Titanium (-AT0) option for peak power efficiency across 100–240V AC and 240V DC inputs.

• Robust +12V/+12.2V rails sized for heavy CPU/GPU loads and always‑on standby power for reliable remote management.

Ideal Use Cases

• Virtualization clusters and VDI at scale.

• High‑performance databases, in‑memory analytics, and big data.

• AI inferencing, model serving, and GPU‑accelerated workloads.

• Software‑defined storage, backup targets, and content delivery.

• Edge and core data centers needing high density, high availability.

Key Highlights at a Glance

• Up to 128 cores/256 threads per system with AMD EPYC 7003/7002.

• 32 DIMM slots, DDR4 up to 3200MT/s; TB‑scale memory support.

• 12x hot‑swap SATA/SAS bays; hybrid SATA+SAS via SAS HBA.

• PCIe Gen4 x16 slots ready for multiple GPUs and high‑speed I/O.

• OCP 3.0/2.0 networking with NCSI for flexible, serviceable connectivity.

• Dual 2000W Platinum or Titanium redundant PSUs for enterprise uptime.

• Ambient support for 280W CPUs at 30°C for demanding deployments.

Accelerate workloads with uncompromising performance, scalable I/O, and rock‑solid reliability—purpose‑built for modern enterprises and service providers.