RAGE UDAY KIRAN

View on GitHub

Computing Environment in RAGE’s lab

1. Physical servers

We have two servers for performing various computing tasks. One server is meant to handle CPU-centric operations, while the other server is meant to handle GPU-centric operations. The details of both the servers are as follows:

  CPU-server GPU-server
Name skytree khalifa
CPU AMD EPYC 7542 32-Core Processor AMD EPYC 7452 32-Core Processor
#NUMA nodes 2 2
#Cores 64 cores (128 virtual cores) 64 cores (128 virtual cores)
RAM 400 GB of main memory 200 GB of main memory
GPUs - two NVidia’s A6000 and Titan RTX
Storage NVME SSDs and Hard drivers
Capacity 8 TB of NVME 14 TB of SSDs and 40 TB of HDDs
Ipaddress 163.143.165.136 163.143.165.137
//: # ( ————- ——————————————————————————- )
//: # ( Name [Calima](https://www.gigabyte.com/jp/Enterprise/Rack-Server/R282-Z94-rev-100) )
//: # ( CPU AMD EPYC 7452 32-Core Processor )
//: # ( #NUMA nodes 2 )
//: # ( #Cores 64 cores (128 virtual cores) )
//: # ( RAM 64 GB of main memory )
//: # ( GPUs NVidia’s 3090 )
//: # ( Storage Hard drivers )
//: # ( Capacity 24 TB of HDDs )
//: # ( Ipaddress 163.143.165.131 )

2. Virtual machines for computing purposes

  1. skytree1.u-aizu.ac.jp (Only CPU)

     163.143.165.136
    
  2. skytree2.u-aizu.ac.jp (Only CPU)

     163.143.165.141
    
  3. khalifa1.u-aizu.ac.jp (CPU + GPU)

     163.143.165.137
    
  4. khalifa2.u-aizu.ac.jp (CPU + GPU)

     163.143.165.138
    
  5. khalifa3.u-aizu.ac.jp (CPU + GPU)

     163.143.165.140