Update images used in nvidia and rocm pipeline testing for 2.16 (back…

…port to releases/2.16.0) (red-hat-data-services#2087)

Update images used in nvidia and rocm pipeline testing for 2.16

Use the workbench images availables in 2.16 RC2

Signed-off-by: Jorge Garcia Oncins <[email protected]>

ods_ci/tests/Resources/Files/pipeline-samples/v2/cache-disabled/gpu/pytorch/pytorch_amd_gpu_availability.py

@@ -3,7 +3,7 @@
 
 #  Runtime: Pytorch with ROCm and Python 3.9 (UBI 9)
 common_base_image = (
-    "quay.io/modh/runtime-images@sha256:a1cfb7bfcff3b2aae2b20b17da83b6683d632403f674a51af6efdfe809a6fc10"
+    "quay.io/modh/runtime-images@sha256:6340efaa92bc54bcede518e890492db626fb9fe96f028c2cd5251f286b2b2852"
 )
 
 
@@ -14,11 +14,9 @@ def add_gpu_toleration(task: PipelineTask, accelerator_type: str, accelerator_li
     kubernetes.add_toleration(task, key=accelerator_type, operator="Exists", effect="NoSchedule")
 
 
-@dsl.component(
-    base_image=common_base_image
-)
+@dsl.component(base_image=common_base_image)
 def verify_gpu_availability(gpu_toleration: bool):
-    import torch
+    import torch  # noqa: PLC0415
 
     cuda_available = torch.cuda.is_available()
     device_count = torch.cuda.device_count()
@@ -30,7 +28,7 @@ def verify_gpu_availability(gpu_toleration: bool):
     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')
+        t = torch.tensor([5, 5, 5], dtype=torch.int64, device="cuda")
     else:
         assert not torch.cuda.is_available()
         assert torch.cuda.device_count() == 0

ods_ci/tests/Resources/Files/pipeline-samples/v2/cache-disabled/gpu/pytorch/pytorch_amd_gpu_availability_compiled.yaml

@@ -42,18 +42,18 @@ deploymentSpec:
 
           '
         - "\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:a1cfb7bfcff3b2aae2b20b17da83b6683d632403f674a51af6efdfe809a6fc10
+          \ *\n\ndef verify_gpu_availability(gpu_toleration: bool):\n    import torch\
+          \  # noqa: PLC0415\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@sha256:6340efaa92bc54bcede518e890492db626fb9fe96f028c2cd5251f286b2b2852
     exec-verify-gpu-availability-2:
       container:
         args:
@@ -80,18 +80,18 @@ deploymentSpec:
 
           '
         - "\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:a1cfb7bfcff3b2aae2b20b17da83b6683d632403f674a51af6efdfe809a6fc10
+          \ *\n\ndef verify_gpu_availability(gpu_toleration: bool):\n    import torch\
+          \  # noqa: PLC0415\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@sha256:6340efaa92bc54bcede518e890492db626fb9fe96f028c2cd5251f286b2b2852
         resources:
           accelerator:
             count: '1'

ods_ci/tests/Resources/Files/pipeline-samples/v2/cache-disabled/gpu/pytorch/pytorch_nvidia_gpu_availability.py

@@ -2,8 +2,9 @@
 from kfp.dsl import PipelineTask
 
 #  Runtime: Pytorch with CUDA and Python 3.9 (UBI 9)
+#  Images for each release can be found here (in the branch for the release)
 common_base_image = (
-    "quay.io/modh/runtime-images@sha256:7d1b065f100666fe46f64a2e8aae888cb41a38b5482bb9b9343b14db05c2a14a"
+    "quay.io/modh/runtime-images@sha256:e1f7ad986f694236a818796af290a99b4e7f73d44cd39ca45860087644d136dd"
 )
 
 
@@ -14,11 +15,9 @@ def add_gpu_toleration(task: PipelineTask, accelerator_type: str, accelerator_li
     kubernetes.add_toleration(task, key=accelerator_type, operator="Exists", effect="NoSchedule")
 
 
-@dsl.component(
-    base_image=common_base_image
-)
+@dsl.component(base_image=common_base_image)
 def verify_gpu_availability(gpu_toleration: bool):
-    import torch
+    import torch  # noqa: PLC0415
 
     cuda_available = torch.cuda.is_available()
     device_count = torch.cuda.device_count()
@@ -30,7 +29,7 @@ def verify_gpu_availability(gpu_toleration: bool):
     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')
+        t = torch.tensor([5, 5, 5], dtype=torch.int64, device="cuda")
     else:
         assert not torch.cuda.is_available()
         assert torch.cuda.device_count() == 0

ods_ci/tests/Resources/Files/pipeline-samples/v2/cache-disabled/gpu/pytorch/pytorch_nvidia_gpu_availability_compiled.yaml

@@ -42,18 +42,18 @@ deploymentSpec:
 
           '
         - "\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@sha256:7d1b065f100666fe46f64a2e8aae888cb41a38b5482bb9b9343b14db05c2a14a
+          \ *\n\ndef verify_gpu_availability(gpu_toleration: bool):\n    import torch\
+          \  # noqa: PLC0415\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:e1f7ad986f694236a818796af290a99b4e7f73d44cd39ca45860087644d136dd
     exec-verify-gpu-availability-2:
       container:
         args:
@@ -80,18 +80,18 @@ deploymentSpec:
 
           '
         - "\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@sha256:7d1b065f100666fe46f64a2e8aae888cb41a38b5482bb9b9343b14db05c2a14a
+          \ *\n\ndef verify_gpu_availability(gpu_toleration: bool):\n    import torch\
+          \  # noqa: PLC0415\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:e1f7ad986f694236a818796af290a99b4e7f73d44cd39ca45860087644d136dd
         resources:
           accelerator:
             count: '1'