Devices supported by HAMi
Applicable version: HAMi v2.11.0. For the historical version, see the v2.10.0 support matrix.
Last verified: 2026-11-06, against the HAMi source, default device configuration, and device guides.
The table below lists device integration status, allocation modes, and multi-card partitioning capabilities for released and experimental HAMi integrations. See the device guides below for the required plugins, drivers, and container runtimes.
Stable means included in an official HAMi release. Release status in this table is checked against v2.11.0. Experimental means the integration is still in development or validation and has not been officially released.
Teco integration is not included in v2.11.0. Device models and capabilities remain to be validated. Track progress in HAMi #3369 and documentation PR #979.
Capability columns:
- Memory control: The memory quota or partition size that a container can request. Templates and fixed partition sizes round memory allocations up to the corresponding size.
- Compute control: Devices control compute resources through time slices, core allocation, hardware partitions, or QoS policies. See the linked references in the table for limitations.
- Multi-card partitioning: A single container can allocate resources across multiple physical devices while using only part of the resources on some of those devices. Allocating only whole devices does not count as multi-card partitioning. See the linked references for allocation granularity and limitations.
N/Ameans the mode allocates whole devices only.
Component integration guides
For NVIDIA sharing, Volcano vGPU and KAI Scheduler retain their own scheduling capabilities and use HAMi-core for runtime isolation; the KAI Scheduler integration provides memory isolation. HAMi-DRA works with the Kubernetes scheduler through DRA drivers; its NVIDIA sharing path uses HAMi-core to enforce memory and compute limits at runtime.
The table below links to device and scheduling component integration guides, including installation steps, configuration requirements, and limitations. - means no guide is available.
| Manufacturer | HAMi | HAMi-DRA | Volcano | KAI-scheduler |
|---|---|---|---|---|
| NVIDIA | Allocate device memory · Remote GPU | Dynamic Resource Allocation | Use Volcano vGPU | Use KAI Scheduler |
| AMD | Enable AMD GPU sharing | - | - | - |
| AWS | Manage AWS Neuron devices | - | - | - |
| Biren | Enable Biren sharing | - | - | - |
| Cambricon | Enable Cambricon MLU sharing | - | - | - |
| Enflame | Enable Enflame GCU sharing | - | - | - |
| Huawei Ascend | Enable Huawei Ascend sharing | - | Volcano Ascend vNPU | - |
| Hygon | Enable Hygon HCU sharing | - | - | - |
| Iluvatar | Enable Iluvatar GPU sharing | - | - | - |
| Kunlunxin | Whole-device scheduling · VXPU sharing | - | - | - |
| MetaX | Enable MetaX GPU sharing | - | - | - |
| Mthreads | Enable Mthreads GPU sharing | - | - | - |
| Teco (Tecorigin) | Integration guide (in development) | - | - | - |
| Vastai | Enable Vastai sharing | - | - | - |
Each linked guide covers device-specific setup, configuration notes, and known limitations. Read it before deploying that device.
