mpi-operator
kserve
Our great sponsors
mpi-operator | kserve | |
---|---|---|
1 | 3 | |
395 | 3,025 | |
1.8% | 6.6% | |
7.3 | 9.4 | |
1 day ago | 2 days ago | |
Go | Python | |
Apache License 2.0 | Apache License 2.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
mpi-operator
-
Scaling Kubernetes to 7,500 Nodes
Hi, kube-scheduler maintainer here, currently looking into enabling MPI use cases in k8s.
We started a discussion in https://github.com/kubeflow/mpi-operator/issues/315
kserve
-
Show HN: Software for Remote GPU-over-IP
Inference servers essentially turn a model running on CPU and/or GPU hardware into a microservice.
Many of them support the kserve API standard[0] that supports everything from model loading/unloading to (of course) inference requests across models, versions, frameworks, etc.
So in the case of Triton[1] you can have any number of different TensorFlow/torch/tensorrt/onnx/etc models, versions, and variants. You can have one or more Triton instances running on hardware with access to local GPUs (for this example). Then you can put standard REST and or grpc load balancers (or whatever you want) in front of them, hit them via another API, whatever.
Now all your applications need to do to perform inference is do an HTTP POST (or use a client[2]) for model input, Triton runs it on a GPU (or CPU if you want), and you get back whatever the model output is.
Not a sales pitch for Triton but it (like some others) can also do things like dynamic batching with QoS parameters, automated model profiling and performance optimization[3], really granular control over resources, response caching, python middleware for application/biz logic, accelerated media processing with Nvidia DALI, all kinds of stuff.
[0] - https://github.com/kserve/kserve
[1] - https://github.com/triton-inference-server/server
[2] - https://github.com/triton-inference-server/client
[3] - https://github.com/triton-inference-server/model_analyzer
-
Run your first Kubeflow pipeline
Kubeflow has multiple components: central dashboard, Kubeflow Notebooks to manage Jupyter notebooks, Kubeflow Pipelines for building and deploying portable, scalable machine learning (ML) workflows based on Docker containers, KF Serving for model serving (apparently superseded by KServe), Katib for hyperparameter tuning and model search, and training operators such as TFJob for training TF models on Kubernetes.
-
[D] Serverless solutions for GPU inference (if there's such a thing)
If you can run on Kubernetes then KFServing is an open source solution that allows for GPU inference and is built upon Knative to allow scale to zero for GPU based inference. From release 0.5 it also has capabilities for multi-model serving as a alpha feature to allow multiple models to share the same server (and via NVIDIA Triton the same GPU).
What are some alternatives?
kube-batch - A batch scheduler of kubernetes for high performance workload, e.g. AI/ML, BigData, HPC
kubeflow - Machine Learning Toolkit for Kubernetes