Skip to main content

Huawei

System Overview​

The Huawei cluser is a supercomputer based on Kunpeng 920 processors, which are based on the ARMv8.1 architecture. It is a system using a Mellanox high performance network interconnect and running CentOS 7.6 as operating system.

The computing elements of this cluster are split into 3 different blocks:

General purpose computing block, which has 16 nodes with the following characteristics:

  • 2 sockets Kunpeng 920 CPU (ARMv8.1) with 64 cores each @ 2.6GHz for a total of 128 cores per node
  • 256 GB of RAM

Dedicated node for AI Training:

  • 4 sockets Kunpeng 920 CPU (ARMv8.1) with 48 cores each @ 2.6GHz for a total of 192 cores per node
  • 1 TB of RAM
  • 8 Ascend 910A (Huawei Accelerators)

Dedicated node for AI Inference:

  • 2 sockets Kunpeng 920 CPU (ARMv8.1) with 64 cores each @ 2.6GHz for a total of 128 cores per node
  • 256 GB of RAM
  • 5 Atlas 300C (Huawei Accelerators, based on IA Ascend 310 processors)

Remember that the BIOS and kernel reserves memory, so the actual total usable RAM that commands like "free" or "lstopo" report will be slighlty lower than the total theoretical amount.

Login Nodes​

You can connect to the Huawei cluster using the follwing public login node. Please note that only incoming connections are allowed in the whole cluster. The login hostname is:

hualogin1.bsc.es