vGPU configuration issues

[V-yw]

HAMI-version: laste

I have one physical card. I'm using HAMI to split 10 vGPUs. Why is it pending when the request count is 9, but successful when it's 7?

apiVersion: v1 kind: Pod metadata: name: gpu-pod spec: containers: - name: ubuntu-container image: ubuntu:2204-base command: ["bash", "-c", "sleep 86400"] resources: limits: nvidia.com/gpu: 9 # 请求 9 个 vGPU nvidia.com/gpumem: 3000

I want to clarify the relationship between .vgpu and slice.