Xiaochang Wu

icon indicating an email [email protected]

icon company @intel

Results 33 comments of Xiaochang Wu

[PIP] Publish python packages in pip/conda channels

#252

[DOC] Add Known Limitations

#270

[Low-Order Moments] Add other stats for low-order moments

Follow-up in upstream: https://github.com/oneapi-src/oneDAL/issues/2006

[Cloud][Databricks] OAP MLLib jar can not be loaded by Databricks runtime

Could I know if Databricks runtime are using K8S as cluster manager?

[DOC] Add GPU-related docs

#270

[Core] NumericTable can't exceed 2G bytes

related to https://github.com/oneapi-src/oneDAL/issues/1476

[K8S] Encounter executors restart issue, when run Hibench Kmeans workload on AWS EKS environment, with Hibench small scale dataset

What version are you testing? The master branch has some dependencies of GPU libraries that I am working to fix in #111 Are you using K8S? KubernetesClusterSchedulerBackend$KubernetesDriverEndpoint: Disabling executor 1....

[K8S] Encounter executors restart issue, when run Hibench Kmeans workload on AWS EKS environment, with Hibench small scale dataset

@zhixingheyi-tian Pls try the latest master branch to see if everything ok. Also @haojinIntel Be notice to use ./build.sh -p spark-3.1.1 to build. There is an additional -p comparing to...

[K8S] Encounter executors restart issue, when run Hibench Kmeans workload on AWS EKS environment, with Hibench small scale dataset

@zhixingheyi-tian From the log, it's not related to memory. oneCCL ranks are not about to connect to the rank 0 which is the executor listening in 192.168.82.1 port 3000. There...

[K8S] Encounter executors restart issue, when run Hibench Kmeans workload on AWS EKS environment, with Hibench small scale dataset

btw. It's encouraged to use build.sh instead of just using mvn to build. Some additional environments need to be set before calling mvn.