Supported Hardware

Table 1 Hardware environment

Type

Configuration Reference

Server (single-node)

  • Atlas 800 training server (model 9000) (full configuration of NPUs)
  • Atlas 800 training server (model 9010) (full configuration of NPUs)
  • Atlas 800 training server (model 9000) (half configuration of NPUs)
  • Atlas 800 training server (model 9010) (half configuration of NPUs)
  • Atlas 800 inference server (model 3000)
  • Atlas 800 inference server (model 3010)
  • Server (with Atlas 300T training cards)
  • Server (with Atlas 300I inference cards)
  • Server (with Atlas 300I Pro inference cards)
  • Server (with Atlas 300I Duo inference cards)
  • Server (with Atlas 300V Pro video analysis cards)
  • Server (with Atlas 300V video analysis cards)
  • Atlas 200I Soc A1 core board

Server (cluster)

Master nodes support multiple categories of servers, such as TaiShan 200 servers (model 2280) and FusionServer Pro 2288H V5.

Worker nodes:

  • Atlas 800 training server (model 9000) (full configuration of NPUs)
  • Atlas 800 training server (model 9010) (full configuration of NPUs)
  • Atlas 800 training server (model 9000) (half configuration of NPUs)
  • Atlas 800 training server (model 9010) (half configuration of NPUs)
  • Atlas 800 inference server (model 3000)
  • Atlas 800 inference server (model 3010)
  • Server (with Atlas 300T training cards)
  • Server (with Atlas 300I inference cards)
  • Server (with Atlas 300I Pro inference cards)
  • Server (with Atlas 300I Duo inference cards)
  • Server (with Atlas 300V Pro video analysis cards)
  • Server (with Atlas 300V video analysis cards)
  • Atlas 200I Soc A1 core board

Storage node: storage server

Memory

Master node memory > 64 GB

Storage

> 1 TB

For details about the drive space plan, see Drive space plan.

Network

  • Out-of-band management (BMC): ≥ 1 Gbit/s
  • In-band management (SSH): ≥ 1 Gbit/s
  • Service plane: ≥ 10 Gbit/s
  • Storage plane: ≥ 25 Gbit/s
  • Parameter plane: 100 Gbit/s
Table 2 Hardware forms supported by Ascend Docker Runtime

Component

Hardware Form

Ascend Docker Runtime

Atlas 200 AI accelerator module (RC)

Atlas 200I SoC A1A200I SoC A1 core board

Atlas 500 AI edge station (model 3000)

Atlas 500 Pro AI edge server (model 3000)

Atlas 300I inference card(model 3000)

Atlas 300I inference card (model 3010)

Atlas 300I Pro inference cardA300I Pro inference card

Atlas 300V Pro video analysis cardA300V Pro video analysis card

Atlas 300I Duo inference cardA300I Duo inference card

Atlas 300V video analysis cardA300V video analysis card

Atlas 300T training card

Atlas 300T training card (model 9000)

Atlas 300T Pro training card (model 9000)

Atlas 800 training server (model 9000)

Atlas 800 training server (model 9010)

Atlas 800 inference server (model 3000)

Atlas 800 inference server (model 3010)

Atlas 900 PoD (model 9000)