Klaus Ma
Klaus Ma
Hi team, There's a kubedag in kubegene for worklfow; as workflow/dag is a common requirement for batch workload, prefer to seperate kubedage to a standalone repo of volcano-sh. If any...
#### What would you like to be added: Enhances preempt action to handle most common cases, and also provides a framework for customized plugins. #### Why is this needed: Currently,...
**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature /area scheduling /priority important-soon **Description**: GPU topology is important to the performance of running tasks, it necessary to imporve both...
There're several enhancements for Volcano to do for GPU scenarios, let's do more investigation on them: - [ ] GPU sharing: https://github.com/NVIDIA/open-gpu-kernel-modules - [ ] GPU & NUMA: https://docs.open-mpi.org/en/v5.0.x/networking/cuda.html#numa-node-issues -...
**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature **What happened**: 1.12 is end of service, it's better to upgrade k8s version to the latest one :)
**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature **What happened**: It's better to follow the practice for group name of k8s-sigs , e.g. `x-k8s.io`.
**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature **What happened**: Currently, we use cmd/deepcopy-gen to generate deepcopy codes; it's better to replace such kind of source code with...
**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature **Description**: Currently, TAS will help several device & workload to get better performance, e.g. NUMA, GPU, IB/SHARP; but most of...
**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature **Description**: Most of AI/HPC workload require job dependency feature; currently, workload will handle the lifecycle of pods, but scheduler does...
**Is this a BUG REPORT or FEATURE REQUEST?**: /kind feature **Description**: Most of batch workload will handle lots of data, so one of major improvement is to avoid data movement....