Accelerator support

Use these pages to understand onboarding, virtualization granularity, validated models, and known limitations for each accelerator family.

📄️NVIDIA

NVIDIA GPU onboarding, virtualization granularity, and limitations.

📄️Ascend

Ascend accelerator onboarding, virtualization granularity, and limitations.

📄️Enflame

Enflame accelerator onboarding, virtualization granularity, and limitations.

📄️T-Head PPU

T-Head PPU onboarding, virtualization granularity, and limitations.

📄️AMD MI series

AMD MI series onboarding, virtualization granularity, and limitations.

📄️MetaX

MetaX accelerator onboarding, virtualization granularity, and limitations.

📄️Kunlunxin

Kunlunxin accelerator onboarding, virtualization granularity, and limitations.

📄️Hygon DCU

Hygon DCU onboarding, virtualization granularity, and limitations.

📄️Iluvatar

Iluvatar accelerator onboarding, virtualization granularity, and limitations.

📄️Cambricon

Cambricon MLU onboarding, virtualization granularity, and limitations.

📄️Moore Threads

Moore Threads accelerator onboarding, virtualization granularity, and limitations.