Wang Zhang

untraceable GPU memory allocation

13

**Describe the bug** When I was testing triton inference server 19.10, GPU memory usage increases when the following two functions are called: 1. cuCtxGetCurrent 2. cuModuleGetFunction It seems when loading...

**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature **Status**: So far FTLib does not support TensorFlow. When adopted in ElasticDL, we take a NumPy NDArray and wrapped it...

**Is this a BUG REPORT or FEATURE REQUEST?**: > Uncomment only one, leave it on its own line: > /kind bug **What happened**: This is an issue I encountered when...

**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature **What happened**: This might not be urgent at this moment. But before we let contributors help with more dl frameworks...

On master branch, while the tensorflow go has adopted version 1.11.0, in the Makefile the libtensorflow is still using version 1.8.0: https://github.com/oracle/graphpipe-go/blob/master/cmd/graphpipe-tf/Makefile#L151

[feature] Add reconciler.v1 package along with controller.v1

8

The common repo offers a `controller.v1` package which is designed for low-level controller mode in kubebuilder. As we are working on a unified controller, it seems a `reconciler.v1` package will...

While the comment for `CleanPodPolicy` and `RestartPolicy` defines what these policies mean, there is no explanation for `CleanPodPolicyAll`, `CleanPodPolicyRunning`, `CleanPodPolicyNone`, `RestartPolicyAlways`, `RestartPolicyOnFailure` and `RestartPolicyNever`

pr 172 breaks training-operator

6

#172 breaks training-operator when installing crd reproduce steps: 1. change go.mod and go.sum ```shell diff --git a/go.mod b/go.mod index 58f089d7..d491f115 100644 --- a/go.mod +++ b/go.mod @@ -5,7 +5,7 @@ go...

Missing methods in reconciler.v1

1

In https://github.com/kubeflow/common/pull/141, methods implemented as `panic("implement me!")` that are aimed to let developers overriding is removed as the pull request is merged. However, when creating a demo reconciler or even...

1. remove duplicated repos for tf-operator, pytorch-operator and mpi-operator 2. add kubeflow/traininig-operator repo to support kubeflow training on kubernetes 1.22+