Skip to content

Commit

Permalink
Replace local image with quay.io image
Browse files Browse the repository at this point in the history
Signed-off-by: Thomas Parnell <[email protected]>
  • Loading branch information
tdoublep committed Jun 6, 2024
1 parent 61e4036 commit f0921cd
Show file tree
Hide file tree
Showing 3 changed files with 4 additions and 20 deletions.
16 changes: 0 additions & 16 deletions docs/SETUP.md
Original file line number Diff line number Diff line change
Expand Up @@ -160,21 +160,5 @@ nvidia-device-plugin-daemonset-xjb5f 1/1 Runnin
nvidia-device-plugin-validator-dm69w 0/1 Completed 0 14s
nvidia-operator-validator-ztpls 1/1 Running 0 114s
```
## Loading a local fmperf image into the cluster node
Currently the fmperf load-tesing docker image is not available on a remote registry. Therefore, we need to build a docker image from the Dockerfile provided in the repo, and load it into the cluster node.
```
docker build -t fmperf-project/fmperf:local .
kind load fmperf-project/fmperf:local --name gpu-test
```
You can verify if this image is correctly loaded into the cluster node:
```
docker exec -it gpu-test-control-plane crictl images | grep local
docker.io/fmperf-project/fmperf local c20f63b5bb19d 992M
```
The cluster is now ready to run the benchmark. As a first try, run the examples/example_vllm.py script.
2 changes: 1 addition & 1 deletion examples/workload_specifications.yml
Original file line number Diff line number Diff line change
Expand Up @@ -4,4 +4,4 @@ min_output_tokens: 50
max_output_tokens: 50
frac_greedy: 1.0
sample_size: 1
image: docker.io/fmperf-project/fmperf:local
image: quay.io/fmperf/fmperf:main
6 changes: 3 additions & 3 deletions fmperf/WorkloadSpecs.py
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@ class WorkloadSpec:
def __init__(
self,
sample_size: int = 10,
image: str = "fmperf-project/fmperf:local",
image: str = "quay.io/fmperf/fmperf:main",
pvc_name: str = None,
overwrite: bool = False,
):
Expand Down Expand Up @@ -58,7 +58,7 @@ def __init__(
max_output_tokens: int = 20,
frac_greedy: float = 0.5,
sample_size: int = 10,
image: str = "fmperf-project/fmperf:local",
image: str = "quay.io/fmperf/fmperf:main",
pvc_name: str = None,
overwrite: bool = False,
):
Expand Down Expand Up @@ -111,7 +111,7 @@ class HeterogeneousWorkloadSpec(WorkloadSpec):
def __init__(
self,
sample_size: int = 10,
image: str = "fmperf-project/fmperf:local",
image: str = "quay.io/fmperf/fmperf:main",
pvc_name: str = None,
overwrite: bool = False,
):
Expand Down

0 comments on commit f0921cd

Please sign in to comment.