跳到主要内容
版本:v1.3.0

Allocate device core and memory resource

Allocate device core and memory to container

To allocate a certain part of device core resource, you need only to assign the mthreads.com/sgpu-memory and mthreads.com/sgpu-core along with the number of cambricon MLUs you requested in the container using mthreads.com/vgpu

apiVersion: v1
kind: Pod
metadata:
name: gpushare-pod-default
spec:
restartPolicy: OnFailure
containers:
- image: core.harbor.zlidc.mthreads.com:30003/mt-ai/lm-qy2:v17-mpc
imagePullPolicy: IfNotPresent
name: gpushare-pod-1
command: ["sleep"]
args: ["100000"]
resources:
limits:
mthreads.com/vgpu: 1
mthreads.com/sgpu-memory: 32
mthreads.com/sgpu-core: 8