Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Automated test to verify AMD GPU's can be used in Data Science Pipelines #1932

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Automated test to verify AMD GPU's can be used in Data Science Pipelines #1932

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
54 changes: 54 additions & 0 deletions ...rces/Files/pipeline-samples/v2/cache-disabled/gpu/pytorch/pytorch_amd_gpu_availability.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,54 @@
from kfp import compiler, dsl, kubernetes
from kfp.dsl import PipelineTask

# Runtime: Pytorch with ROCm and Python 3.9 (UBI 9)
common_base_image = (
"quay.io/modh/runtime-images@sha256:76d2a5d9ca0bb35c4d9962a7ee6798672afd3a5c7d226d87ef65f952e7cb93cf"
)


def add_gpu_toleration(task: PipelineTask, accelerator_type: str, accelerator_limit: int):
bdattoma marked this conversation as resolved.
Show resolved Hide resolved
print(f"Adding GPU tolerations: {accelerator_type}({accelerator_limit})")
task.set_accelerator_type(accelerator=accelerator_type)
task.set_accelerator_limit(accelerator_limit)
kubernetes.add_toleration(task, key=accelerator_type, operator="Exists", effect="NoSchedule")


@dsl.component(
base_image=common_base_image
)
def verify_gpu_availability(gpu_toleration: bool):
import torch

cuda_available = torch.cuda.is_available()
device_count = torch.cuda.device_count()
print("------------------------------")
print("GPU availability")
print("------------------------------")
print(f"cuda available: {cuda_available}")
print(f"device count: {device_count}")
if gpu_toleration:
assert torch.cuda.is_available()
assert torch.cuda.device_count() > 0
t = torch.tensor([5, 5, 5], dtype=torch.int64, device='cuda')
else:
assert not torch.cuda.is_available()
assert torch.cuda.device_count() == 0
t = torch.tensor([5, 5, 5], dtype=torch.int64)
print(f"tensor: {t}")
print("GPU availability test: PASS")


@dsl.pipeline(
name="pytorch-amd-gpu-availability",
description="Verifies pipeline tasks run on GPU nodes only when tolerations are added",
)
def pytorch_amd_gpu_availability():
verify_gpu_availability(gpu_toleration=False).set_caching_options(False)

task_with_toleration = verify_gpu_availability(gpu_toleration=True).set_caching_options(False)
add_gpu_toleration(task_with_toleration, "amd.com/gpu", 1)


if __name__ == "__main__":
compiler.Compiler().compile(pytorch_amd_gpu_availability, package_path=__file__.replace(".py", "_compiled.yaml"))
139 changes: 139 additions & 0 deletions ...pipeline-samples/v2/cache-disabled/gpu/pytorch/pytorch_amd_gpu_availability_compiled.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,139 @@
# PIPELINE DEFINITION
# Name: pytorch-amd-gpu-availability
# Description: Verifies pipeline tasks run on GPU nodes only when tolerations are added
components:
comp-verify-gpu-availability:
executorLabel: exec-verify-gpu-availability
inputDefinitions:
parameters:
gpu_toleration:
parameterType: BOOLEAN
comp-verify-gpu-availability-2:
executorLabel: exec-verify-gpu-availability-2
inputDefinitions:
parameters:
gpu_toleration:
parameterType: BOOLEAN
deploymentSpec:
executors:
exec-verify-gpu-availability:
container:
args:
- --executor_input
- '{{$}}'
- --function_to_execute
- verify_gpu_availability
command:
- sh
- -c
- "\nif ! [ -x \"$(command -v pip)\" ]; then\n python3 -m ensurepip ||\
\ python3 -m ensurepip --user || apt-get install python3-pip\nfi\n\nPIP_DISABLE_PIP_VERSION_CHECK=1\
\ python3 -m pip install --quiet --no-warn-script-location 'kfp==2.9.0'\
\ '--no-deps' 'typing-extensions>=3.7.4,<5; python_version<\"3.9\"' && \"\
$0\" \"$@\"\n"
- sh
- -ec
- 'program_path=$(mktemp -d)
printf "%s" "$0" > "$program_path/ephemeral_component.py"
_KFP_RUNTIME=true python3 -m kfp.dsl.executor_main --component_module_path "$program_path/ephemeral_component.py" "$@"
'
- "\nimport kfp\nfrom kfp import dsl\nfrom kfp.dsl import *\nfrom typing import\
\ *\n\ndef verify_gpu_availability(gpu_toleration: bool):\n import torch\n\
\n cuda_available = torch.cuda.is_available()\n device_count = torch.cuda.device_count()\n\
\ print(\"------------------------------\")\n print(\"GPU availability\"\
)\n print(\"------------------------------\")\n print(f\"cuda available:\
\ {cuda_available}\")\n print(f\"device count: {device_count}\")\n \
\ if gpu_toleration:\n assert torch.cuda.is_available()\n \
\ assert torch.cuda.device_count() > 0\n t = torch.tensor([5, 5,\
\ 5], dtype=torch.int64, device='cuda')\n else:\n assert not torch.cuda.is_available()\n\
\ assert torch.cuda.device_count() == 0\n t = torch.tensor([5,\
\ 5, 5], dtype=torch.int64)\n print(f\"tensor: {t}\")\n print(\"GPU\
\ availability test: PASS\")\n\n"
image: quay.io/modh/runtime-images@sha256:76d2a5d9ca0bb35c4d9962a7ee6798672afd3a5c7d226d87ef65f952e7cb93cf
exec-verify-gpu-availability-2:
container:
args:
- --executor_input
- '{{$}}'
- --function_to_execute
- verify_gpu_availability
command:
- sh
- -c
- "\nif ! [ -x \"$(command -v pip)\" ]; then\n python3 -m ensurepip ||\
\ python3 -m ensurepip --user || apt-get install python3-pip\nfi\n\nPIP_DISABLE_PIP_VERSION_CHECK=1\
\ python3 -m pip install --quiet --no-warn-script-location 'kfp==2.9.0'\
\ '--no-deps' 'typing-extensions>=3.7.4,<5; python_version<\"3.9\"' && \"\
$0\" \"$@\"\n"
- sh
- -ec
- 'program_path=$(mktemp -d)
printf "%s" "$0" > "$program_path/ephemeral_component.py"
_KFP_RUNTIME=true python3 -m kfp.dsl.executor_main --component_module_path "$program_path/ephemeral_component.py" "$@"
'
- "\nimport kfp\nfrom kfp import dsl\nfrom kfp.dsl import *\nfrom typing import\
\ *\n\ndef verify_gpu_availability(gpu_toleration: bool):\n import torch\n\
\n cuda_available = torch.cuda.is_available()\n device_count = torch.cuda.device_count()\n\
\ print(\"------------------------------\")\n print(\"GPU availability\"\
)\n print(\"------------------------------\")\n print(f\"cuda available:\
\ {cuda_available}\")\n print(f\"device count: {device_count}\")\n \
\ if gpu_toleration:\n assert torch.cuda.is_available()\n \
\ assert torch.cuda.device_count() > 0\n t = torch.tensor([5, 5,\
\ 5], dtype=torch.int64, device='cuda')\n else:\n assert not torch.cuda.is_available()\n\
\ assert torch.cuda.device_count() == 0\n t = torch.tensor([5,\
\ 5, 5], dtype=torch.int64)\n print(f\"tensor: {t}\")\n print(\"GPU\
\ availability test: PASS\")\n\n"
image: quay.io/modh/runtime-images@sha256:76d2a5d9ca0bb35c4d9962a7ee6798672afd3a5c7d226d87ef65f952e7cb93cf
resources:
accelerator:
count: '1'
type: amd.com/gpu
pipelineInfo:
description: Verifies pipeline tasks run on GPU nodes only when tolerations are
added
name: pytorch-amd-gpu-availability
root:
dag:
tasks:
verify-gpu-availability:
cachingOptions: {}
componentRef:
name: comp-verify-gpu-availability
inputs:
parameters:
gpu_toleration:
runtimeValue:
constant: false
taskInfo:
name: verify-gpu-availability
verify-gpu-availability-2:
cachingOptions: {}
componentRef:
name: comp-verify-gpu-availability-2
inputs:
parameters:
gpu_toleration:
runtimeValue:
constant: true
taskInfo:
name: verify-gpu-availability-2
schemaVersion: 2.1.0
sdkVersion: kfp-2.9.0
---
platforms:
kubernetes:
deploymentSpec:
executors:
exec-verify-gpu-availability-2:
tolerations:
- effect: NoSchedule
key: amd.com/gpu
operator: Exists
54 changes: 54 additions & 0 deletions ...s/Files/pipeline-samples/v2/cache-disabled/gpu/pytorch/pytorch_nvidia_gpu_availability.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,54 @@
from kfp import compiler, dsl, kubernetes
from kfp.dsl import PipelineTask

# Runtime: Pytorch with CUDA and Python 3.9 (UBI 9)
common_base_image = (
"quay.io/modh/runtime-images@sha256:cee154f6db15de27929362f91baa128fc4f79b9c1930ab0f27561174d39aadfa"
)


def add_gpu_toleration(task: PipelineTask, accelerator_type: str, accelerator_limit: int):
print(f"Adding GPU tolerations: {accelerator_type}({accelerator_limit})")
task.set_accelerator_type(accelerator=accelerator_type)
task.set_accelerator_limit(accelerator_limit)
kubernetes.add_toleration(task, key=accelerator_type, operator="Exists", effect="NoSchedule")


@dsl.component(
base_image=common_base_image
)
def verify_gpu_availability(gpu_toleration: bool):
import torch

cuda_available = torch.cuda.is_available()
device_count = torch.cuda.device_count()
print("------------------------------")
print("GPU availability")
print("------------------------------")
print(f"cuda available: {cuda_available}")
print(f"device count: {device_count}")
if gpu_toleration:
assert torch.cuda.is_available()
assert torch.cuda.device_count() > 0
t = torch.tensor([5, 5, 5], dtype=torch.int64, device='cuda')
else:
assert not torch.cuda.is_available()
assert torch.cuda.device_count() == 0
t = torch.tensor([5, 5, 5], dtype=torch.int64)
print(f"tensor: {t}")
print("GPU availability test: PASS")


@dsl.pipeline(
name="pytorch-nvidia-gpu-availability",
description="Verifies pipeline tasks run on GPU nodes only when tolerations are added",
)
def pytorch_nvidia_gpu_availability():
verify_gpu_availability(gpu_toleration=False).set_caching_options(False)

task_with_toleration = verify_gpu_availability(gpu_toleration=True).set_caching_options(False)
add_gpu_toleration(task_with_toleration, "nvidia.com/gpu", 1)


if __name__ == "__main__":
compiler.Compiler().compile(pytorch_nvidia_gpu_availability, package_path=__file__.replace(".py", "_compiled.yaml"))
139 changes: 139 additions & 0 deletions ...eline-samples/v2/cache-disabled/gpu/pytorch/pytorch_nvidia_gpu_availability_compiled.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,139 @@
# PIPELINE DEFINITION
# Name: pytorch-nvidia-gpu-availability
# Description: Verifies pipeline tasks run on GPU nodes only when tolerations are added
components:
comp-verify-gpu-availability:
executorLabel: exec-verify-gpu-availability
inputDefinitions:
parameters:
gpu_toleration:
parameterType: BOOLEAN
comp-verify-gpu-availability-2:
executorLabel: exec-verify-gpu-availability-2
inputDefinitions:
parameters:
gpu_toleration:
parameterType: BOOLEAN
deploymentSpec:
executors:
exec-verify-gpu-availability:
container:
args:
- --executor_input
- '{{$}}'
- --function_to_execute
- verify_gpu_availability
command:
- sh
- -c
- "\nif ! [ -x \"$(command -v pip)\" ]; then\n python3 -m ensurepip ||\
\ python3 -m ensurepip --user || apt-get install python3-pip\nfi\n\nPIP_DISABLE_PIP_VERSION_CHECK=1\
\ python3 -m pip install --quiet --no-warn-script-location 'kfp==2.9.0'\
\ '--no-deps' 'typing-extensions>=3.7.4,<5; python_version<\"3.9\"' && \"\
$0\" \"$@\"\n"
- sh
- -ec
- 'program_path=$(mktemp -d)
printf "%s" "$0" > "$program_path/ephemeral_component.py"
_KFP_RUNTIME=true python3 -m kfp.dsl.executor_main --component_module_path "$program_path/ephemeral_component.py" "$@"
'
- "\nimport kfp\nfrom kfp import dsl\nfrom kfp.dsl import *\nfrom typing import\
\ *\n\ndef verify_gpu_availability(gpu_toleration: bool):\n import torch\n\
\n cuda_available = torch.cuda.is_available()\n device_count = torch.cuda.device_count()\n\
\ print(\"------------------------------\")\n print(\"GPU availability\"\
)\n print(\"------------------------------\")\n print(f\"cuda available:\
\ {cuda_available}\")\n print(f\"device count: {device_count}\")\n \
\ if gpu_toleration:\n assert torch.cuda.is_available()\n \
\ assert torch.cuda.device_count() > 0\n t = torch.tensor([5, 5,\
\ 5], dtype=torch.int64, device='cuda')\n else:\n assert not torch.cuda.is_available()\n\
\ assert torch.cuda.device_count() == 0\n t = torch.tensor([5,\
\ 5, 5], dtype=torch.int64)\n print(f\"tensor: {t}\")\n print(\"GPU\
\ availability test: PASS\")\n\n"
image: quay.io/modh/runtime-images@sha256:cee154f6db15de27929362f91baa128fc4f79b9c1930ab0f27561174d39aadfa
exec-verify-gpu-availability-2:
container:
args:
- --executor_input
- '{{$}}'
- --function_to_execute
- verify_gpu_availability
command:
- sh
- -c
- "\nif ! [ -x \"$(command -v pip)\" ]; then\n python3 -m ensurepip ||\
\ python3 -m ensurepip --user || apt-get install python3-pip\nfi\n\nPIP_DISABLE_PIP_VERSION_CHECK=1\
\ python3 -m pip install --quiet --no-warn-script-location 'kfp==2.9.0'\
\ '--no-deps' 'typing-extensions>=3.7.4,<5; python_version<\"3.9\"' && \"\
$0\" \"$@\"\n"
- sh
- -ec
- 'program_path=$(mktemp -d)
printf "%s" "$0" > "$program_path/ephemeral_component.py"
_KFP_RUNTIME=true python3 -m kfp.dsl.executor_main --component_module_path "$program_path/ephemeral_component.py" "$@"
'
- "\nimport kfp\nfrom kfp import dsl\nfrom kfp.dsl import *\nfrom typing import\
\ *\n\ndef verify_gpu_availability(gpu_toleration: bool):\n import torch\n\
\n cuda_available = torch.cuda.is_available()\n device_count = torch.cuda.device_count()\n\
\ print(\"------------------------------\")\n print(\"GPU availability\"\
)\n print(\"------------------------------\")\n print(f\"cuda available:\
\ {cuda_available}\")\n print(f\"device count: {device_count}\")\n \
\ if gpu_toleration:\n assert torch.cuda.is_available()\n \
\ assert torch.cuda.device_count() > 0\n t = torch.tensor([5, 5,\
\ 5], dtype=torch.int64, device='cuda')\n else:\n assert not torch.cuda.is_available()\n\
\ assert torch.cuda.device_count() == 0\n t = torch.tensor([5,\
\ 5, 5], dtype=torch.int64)\n print(f\"tensor: {t}\")\n print(\"GPU\
\ availability test: PASS\")\n\n"
image: quay.io/modh/runtime-images@sha256:cee154f6db15de27929362f91baa128fc4f79b9c1930ab0f27561174d39aadfa
resources:
accelerator:
count: '1'
type: nvidia.com/gpu
pipelineInfo:
description: Verifies pipeline tasks run on GPU nodes only when tolerations are
added
name: pytorch-nvidia-gpu-availability
root:
dag:
tasks:
verify-gpu-availability:
cachingOptions: {}
componentRef:
name: comp-verify-gpu-availability
inputs:
parameters:
gpu_toleration:
runtimeValue:
constant: false
taskInfo:
name: verify-gpu-availability
verify-gpu-availability-2:
cachingOptions: {}
componentRef:
name: comp-verify-gpu-availability-2
inputs:
parameters:
gpu_toleration:
runtimeValue:
constant: true
taskInfo:
name: verify-gpu-availability-2
schemaVersion: 2.1.0
sdkVersion: kfp-2.9.0
---
platforms:
kubernetes:
deploymentSpec:
executors:
exec-verify-gpu-availability-2:
tolerations:
- effect: NoSchedule
key: nvidia.com/gpu
operator: Exists
Loading
Loading