forked from mlcommons/inference_results_v2.0
/
mlperf_inferencev20.yaml
38 lines (38 loc) · 1.11 KB
/
mlperf_inferencev20.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
apiVersion: v1
kind: Pod
metadata:
name: mlinference
namespace: test
spec:
restartPolicy: OnFailure
containers:
- name: inference
image: "quay.io/dfeddema/mlperfinferencev20:1.0.0"
command: ["/bin/bash", "-ec","cd /work;export MLPERF_SCRATCH_PATH=/data/scratch_space; make run RUN_ARGS='--benchmarks=3d-unet,bert,rnnt,resnet50,ssd-mobilenet,ssd-resnet34 --scenarios=offline,singlestream --fast'" ]
runAsUser: 1000
env:
- name: NVIDIA_VISIBLE_DEVICES
value: all
- name: NVIDIA_DRIVER_CAPABILITIES
value: "compute,utility"
- name: NVIDIA_REQUIRE_CUDA
value: "cuda>=5.0"
resources:
limits:
nvidia.com/gpu: 1 # requesting 1 GPU
securityContext:
privileged: true
volumeMounts:
- mountPath: /work
name: test-volume
- mountPath: /data
name: test-volume2
volumes:
- name: test-volume
hostPath:
# directory location on host
path: /root/inference_results_v2.0/closed/NVIDIA
- name: test-volume2
hostPath:
# directory location on host
path: /data