HCCL_INTRA_ROCE_ENABLE

Description

Specifies whether to use the RoCE path for multi-processor communication on a server.

This environment variable can be configured independently or used together with HCCL_INTRA_PCIE_ENABLE. The configuration descriptions are as follows:

• If HCCL_INTRA_PCIE_ENABLE is set to 1 and HCCL_INTRA_ROCE_ENABLE is not set, the PCIe loop is used for communication between multiple devices on a server.
• If HCCL_INTRA_PCIE_ENABLE is not set and HCCL_INTRA_ROCE_ENABLE is set to 1, the RoCE loop is used for communication between multiple devices on a server.
• If HCCL_INTRA_PCIE_ENABLE is set to 1 and HCCL_INTRA_ROCE_ENABLE is set to 0, the PCIe loop is used for communication between multiple devices on a server.
• If HCCL_INTRA_PCIE_ENABLE is set to 0 and HCCL_INTRA_ROCE_ENABLE is set to 1, the RoCE loop is used for communication between multiple devices on a server.
• If HCCL_INTRA_PCIE_ENABLE is set to 0 and HCCL_INTRA_ROCE_ENABLE is set to 0, the PCIe loop is used for communication between multiple devices on a server.
  

  • If neither HCCL_INTRA_PCIE_ENABLE nor HCCL_INTRA_ROCE_ENABLE is configured, the PCIe loop is used for communication between multiple devices on a server by default.
  • HCCL_INTRA_PCIE_ENABLE and HCCL_INTRA_ROCE_ENABLE cannot be set to 1 at the same time.

Example

export HCCL_INTRA_ROCE_ENABLE=1

Applicability

Atlas Training Series Product: Only the Atlas 300T Pro training card is supported.