AceleMax DGS-1

4U 8x Tesla SXM2 GPU Server

 

  • Deep Learning Training and Inference
  • Delivers up to 5,120 Tensor cores, 40,960 CUDA Cores, 62.4 Teraflops double precision, 125.6 Teraflops single precision, and 1,000 Teraflops Tensor performance.

Request a Quote

Reference # 4B28205

Features

Powered by the latest NVIDIA V100 GPU accelerator, the world’s fastest, most advanced, and most efficient data center GPUs ever built, delivering performance of up to 100 CPUs in a single Tesla V100 GPU, enabling data scientists, researchers, and engineers to tackle challenges that were once thought impossible. The DGS-1 delivers up to 5,120 Tensor cores, 40,960 CUDA Cores, 62.4 Teraflops double precision, 125.6 Teraflops single precision, and 1,000 Teraflops Tensor performance.

Applications

  • Hyperscale Datacenters, Supercomputing Centers, Consumer Internet Companies, Higher Ed Research, Government, Healthcare, Financial Services, Retail, Manufacturing

Processor

Supports two 2nd Generation Intel® Xeon® Processor Scalable Family series, LGA 3647 Socket P0, up to 205W TDP

Chipset

  • Intel® C621 chipset

GPU Support & Quantity

  • Supports eight Nvidia V100 SXM2 GPU cards in a 4U chassis
  • Delivers up to ,120 Tensor cores, 40,960 CUDA Cores, 62.4 Teraflops double precision, 125.6 Teraflops single precision, and 1,000 Teraflops Tensor performance
  • Up to 256GB dedicated HBM2 GPU memory (organized 32GB per V100 SXM2 GPU)
  • Supports up to 300GB/s GPU interconnect bandwidth with NVLink technology
  • Delivers the lowest latency of data access and transfer

System Memory (Maximum)

  • 24 DDR4 DIMM slots
  • Up to 6TB 3DS ECC DDR4-2933MHz RDIMM/LRDIMM
  • Supports Intel® Optane™ DCPMM

Expansion Slots

  • 4 x PCI-E 3.0 x16 (LP, GPU tray for GPUDirect RDMA) slots
  • 2 x PCI-E 3.0 x16 (LP, CPU tray) slots or up to 8x 2.5″ NVMe Storage Devices

Connectivity

  • On-board ASPEED AST2500 BM
  • Dual Port 10GBase-T from Intel X540 Ethernet Controller
  • 1 x RJ45 Dedicated IPMI LAN port

Management

  • Support for Intelligent Platform Management Interface v.2.0
  • IPMI 2.0 with virtual media over LAN and KVM-over-LAN support

Drive Bays

  • 16-port 2U SAS3 12Gbps hybrid backplane, support up to 8x 2.5″ SAS3/SATA3 HDD/SSD
  • 8x SAS3/SATA3/NVMe Storage Devices

Power Supply

2200W Redundant Titanium Level Power Supplies with PMBus

System Dimensions

17.6” x 7” x 31.7” / 447mm x 178mm x 805mm

Optimized for Turnkey Solutions

Enable powerful design, training, and visualization with built-in software tools including TensorFlow, Caffe, Torch, Theano, BIDMach cuDNN, NVIDIA CUDA Toolkit and NVIDIA DIGITS.