advanced_hpc_logo
phone
request_a_quote
 
     
   
  greenbulletRackmount Servers  
bar
  
  greenbulletHigh Performance Computing  
bar
  
  greenbulletStorage Servers  
bar
  
  greenbulletRAID Storage  
bar
  
  greenbulletTower Servers  
bar
  
  greenbullet GPU Computing  
    bulletwhite GPU Blade  
      GPU Processor Blade  
    bulletwhite8-GPU Rackmount Servers  
    bulletwhiteTesla S-Series  
    bulletwhiteGPU Appliance  
    bulletwhiteTower/4U Rackmount Servers  
    bulletwhiteAMD GPU-optimized Servers  
    bulletwhiteIntel GPU-optimized Servers  
    bulletwhiteGPU Software  
bar
  
  greenbulletNetwork Attached Storage  
bar
  
  greenbulletBackup  
bar
  
  greenbulletSupport  
bar
  
  greenbulletCompany Information  
bar
  
  greenbulletGovernment & Education  
   
   
 
   Saturn GPU Blade
 
   

Advanced HPC's GPU Blade is an unprecedented high-density GPU blade which supports two NVIDIA Tesla M2050/M2070/M2090 GPU cards with non-blocking native Gen2 PCI-Express x16 connections per node.  Compared to other GPU-enabled blade solutions, our GPU Blade provides more than double the number of GPUs per 1U of rack space, and it is perfect for high-performance computing (HPC) applications in fields like medical imaging, oil and gas exploration, quantum chemistry, financial simulation, genomics and astrophysics. AHPC's new GPU Blade system enables up to 120 of the highest performing Fermi GPUs plus 120 of the latest Intel® Xeon® 5600 series processors to be deployed in a standard 42U rack. Equipped with dual IOH chips and dual 40Gb/s QDR IB HCA per blade, this GPU SuperBlade solution delivers maximum performance with the best CPU to GPU balance and optimized I/O.

  
   

GPU103-SX33

 

 

Features

  • Up to 20x Tesla 20-Series M2050/M2070/M2090 GPUs
  • Up to 20x Intel Xeon 5600 series CPUs
  • Up to 120GB DDR5 GPU memory
  • Up to 960GB 1333/1066/800MHz ECC Reg CPU memory (96GB max per host server)
  • 1x SATA DOM
  • 1x USB flash drive
  • Dual 40Gb InfiniBand, 10Gb Ethernet or 8Gb FCoE supported via optional mezzanine card
  • Dual-port GbE NIC
  • Redundant GBX connectors
  • Dual-port Gigabit Ethernet controller per blade
  • Onboard BMC for IPMI 2.0 support – KVM over IP, remote Virtual Media, etc.
  • Up to 10,240 CUDA cores, 26.62 teraflops of single precision and 13.30 teraflops of double precision performance in 7U
  • Up to 61,440 CUDA cores, 159.72 teraflops of single precision and 79.80 teraflops of double precision performance in a standard 42U cabinet
  • Up to four hot-swap 1400W / 2000W power supplies, N+1 redundant
  • Standard 3 year warranty, Optional advance replacement and on-site service warranty available
 
intel
 
   For detailed specifications on the GPU Processor Blade, click here. SBI-7126TG
 
 
   

Home | Products | Support | Company | News | Partners | Privacy | Contact Us

Last updated 06/02/2011