1

export IMAGE=ascendai/pytorch:2.1.0

 1
 2
 3
 4
 5
 6
 7
 8
 9
10

nerdctl run --security-opt apparmor=unconfined --security-opt seccomp=unconfined --rm -it --ipc=host \
                     --device=/dev/davinci7 \
                     --device=/dev/davinci_manager \
                     --device=/dev/devmm_svm \
                     --device=/dev/hisi_hdc \
                     -v /usr/local/Ascend/driver:/usr/local/Ascend/driver \
                     -v /usr/local/Ascend/add-ons/:/usr/local/Ascend/add-ons/ \
                     -v /usr/local/sbin/npu-smi:/usr/local/sbin/npu-smi \
                     ${IMAGE} \
                     /bin/bash

1
2

export IMAGE=ascendai/pytorch:2.1.0
export NodeName=

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20

kubectl create -f - <<EOF
apiVersion: v1
kind: Pod
metadata:
  name: test-ascend-pod
  namespace: default
spec:
  restartPolicy: Never
  nodeName: ${NodeName}
  containers:
  - name: ascend-container
    image: ${IMAGE}
    command: ["/bin/bash"]
    args: ["-c", "sleep infinity"]
    resources:
      limits:
        huawei.com/Ascend910: "1"
      requests:
        huawei.com/Ascend910: "1"
EOF

1
2
3
4
5
6
7
8
9

mkdir -p /data/paascontainer/ops
cd /data/paascontainer/ops
rm -f *.run
wget http://mirrors-internal.cmecloud.cn/coca/huawei/910b/Ascend-hdk-910b-npu-firmware_7.7.0.1.231.run
wget http://mirrors-internal.cmecloud.cn/coca/huawei/910b/Ascend-hdk-910b-npu-driver_25.0.rc1.1_linux-aarch64.run
chmod +x *.run
./Ascend-hdk-*-npu-firmware_*.run --upgrade
./Ascend-hdk-*-npu-driver_*.run --upgrade
reboot

1

DeviceInfoCfg: '{"DeviceInfo":{"DeviceList":{"huawei.com/Ascend910":"Ascend910-10,Ascend910-11,Ascend910-12,Ascend910-13,Ascend910-14,Ascend910-8,Ascend910-9","huawei.com/Ascend910-Fault":"[{\"fault_type\":\"CardUnhealthy\",\"npu_name\":\"Ascend910-15\",\"large_model_fault_level\":\"ManuallySeparateNPU\",\"fault_level\":\"ManuallySeparateNPU\",\"fault_handling\":\"ManuallySeparateNPU\",\"fault_code\":\"\",\"fault_time_and_level_map\":{}}]","huawei.com/Ascend910-NetworkUnhealthy":"","huawei.com/Ascend910-Recovering":"","huawei.com/Ascend910-Unhealthy":"Ascend910-15"}},"SuperPodID":-1,"ServerIndex":1023}'

1

kubectl -n kube-system delete cm mindx-dl-deviceinfo-