Device supported by HAMi

The table below lists the devices supported by HAMi. This matrix reflects HAMi v2.10.0, the latest released version.

Type	Manufacturer	Models	Status	MemoryIsolation	CoreIsolation	MultiCard Partitioning
GPU	NVIDIA	All	Stable	Yes	Yes	Yes
GPU	AMD	Instinct / ROCm	Stable	Yes	Yes	No
Neuron	AWS	Inf, Trn	Stable	No	Yes	No
GPU	Biren	Biren166M	Stable	No	No	No
MLU	Cambricon	370, 590	Stable	Yes	Yes	No
GCU	Enflame	S60	Stable	Yes	Yes	No
NPU	Huawei Ascend	910B, 910B3, 910C, 310P	Stable	Yes	Yes	No
HCU	Hygon	All	Stable	Yes	Yes	No
GPU	Iluvatar	All	Stable	Yes	Yes	No
XPU	Kunlunxin	P800	Stable	Yes	No	No
GPU	MetaX	MXC500	Stable	Yes	Yes	No
GPU	Mthreads	MTT S4000	Stable	Yes	Yes	No
DPU	Teco	Checking	Under Validation	No	No	No
GPU	Vastai	VA16	Stable	No	No	No

Support status:

• Stable - Available in the latest released HAMi version.
• Experimental - Implemented in HAMi but not yet included in a released version.
• Under Validation - Support is still being implemented; not yet functional.

Capability columns:

• MemoryIsolation - Whether HAMi enforces a hard VRAM limit per container: workloads that exceed their requested memory are rejected instead of drawing on the full physical device memory.
• CoreIsolation - Whether HAMi enforces a hard compute usage limit per container: kernel execution is throttled to stay within the requested share instead of using the physical device's compute freely.
• MultiCard Partitioning - Whether a single Pod can use partitioned memory or compute resources across multiple physical cards. No means multiple whole cards can still be requested, but memory or core resources must not be specified when requesting more than one card.

Which component supports which device

HAMi is one of four ways to reach these devices. The other three schedule the workload themselves and rely on HAMi-core for isolation.

Each cell links to the guide for that device and component. A dash means there is no guide today, not that the combination is impossible.

Manufacturer	HAMi	HAMi-DRA	Volcano	KAI-scheduler
NVIDIA	Allocate device memory	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	Enable Kunlunxin scheduling	-	-	-
MetaX	Enable MetaX GPU sharing	-	-	-
Mthreads	Enable Mthreads GPU sharing	-	-	-
Teco	-	-	-	-
Vastai	Enable Vastai sharing	-	-	-

Each linked guide covers device-specific setup, configuration notes, and known limitations. Read it before deploying that device.

Every page, by vendor

All vendor pages are listed here so this page reaches each of them.

NVIDIA

• Allocate device memory to container
• Dynamic Resource Allocation
• Enable dynamic MIG feature
• Scheduling Policies
• Allocate device core to container
• Assign to certain device type
• Assign to Certain Device UUID
• Using Extended ResourceQuota for NVIDIA Devices
• Use Exclusive GPU
• Allocate certain device memory to container
• Allocate a part of device memory by percentage to container
• Allocate device core to container
• Assign task to a certain type
• Assign task to a certain GPU
• Assign task to MIG instance

AMD

• Enable AMD GPU Sharing
• How AMD GPU sharing works
• Configure the AMD device plugin
• Troubleshoot AMD GPU sharing
• AMD GPU test status
• Allocate a whole GPU
• Allocate device memory
• Allocate device memory by percentage
• Allocate device core and memory resource
• Assign task to a certain GPU type
• Avoid a certain GPU type
• Assign task to a certain GPU by UUID
• Avoid a certain GPU by UUID
• Allocate GPUs on one NUMA node
• Allocate XGMI-linked GPUs
• Choose a GPU scheduling policy
• Set the queue priority of a pod

AWS Neuron

• Enable AWS-Neuron device Sharing
• Allocate AWS Neuron core
• Allocate AWS Neuron device

Biren

• Enable Biren Sharing
• Allocate Biren Device

Cambricon

• Enable Cambricon MLU Sharing
• Allocate device memory
• Allocate device core usage
• Assign to certain device type
• Allocate device core and memory to container
• Allocate exclusive device

Enflame

• Enable Enflame GCU Sharing

Huawei Ascend

• Enable Huawei Ascend sharing
• Huawei Ascend device template
• Allocate 310P slice
• Allocate 910B slice
• Allocate exclusive device
• Soft slicing (hami-vnpu-core)

Hygon

• Enable Hygon HCU sharing
• Allocate device memory
• Allocate device core to container
• Assign to certain device
• Allocate device core and memory resource
• Allocate exclusive device
• Assign task to certain HCU cards

Iluvatar

• Enable Iluvatar GPU Sharing
• Allocate BI-V150 slice
• Allocate MR-V100 slice
• Allocate exclusive BI-V150 device
• Allocate exclusive MR-V100 device

Kunlunxin

• Enable Kunlunxin GPU Topology-Aware Scheduling
• Enable Kunlunxin VXPU
• Allocate a whole xpu card
• Allocate vxpu device

MetaX

• Enable MetaX GPU sharing
• Allocate device core and memory resource
• Allocate exclusive device
• Allocate specific QoS policy devices
• Enable MetaX GPU topology-aware scheduling
• Binpack schedule policy
• Spread schedule policy
• Allocate metax device
• Binpack schedule policy
• Spread schedule policy

Mthreads

• Enable Mthreads GPU sharing
• Allocate device memory to container
• Allocate device core to container
• Allocate device core and memory resource
• Allocate exclusive device

Vastai

• Enable Vastai Sharing
• Allocate Vastai Device