Home > Workload Solutions > High Performance Computing > White Papers > Machine Learning Using Red Hat OpenShift Container Platform > Recommended configuration for OCP with Kubeflow
Dell EMC recommends the following modifications to OpenShift Container Platform v3.11 to host ML workloads using Kubeflow.
ML/DL training is one of the most computationally intensive workloads in the enterprise data center. Dell EMC recommends using the latest Intel Xeon Scalable processor family to take advantage of the Intel Deep Learning Boost optimizations that are targeted for complex ML workloads.
The following table shows the recommended configuration for application nodes:
Table 1. Application node configuration
Hardware |
Description |
SKU |
Server |
3 x Dell EMC PowerEdge R640 |
210-AKWU |
CPU |
2 x Intel Xeon Gold 6248 processor |
338-BRVO
|
Memory |
384 GB (12 x 32 GB 2666MHz DDR4 ECC RDIMM) |
370-ADNF |
Storage |
Capacity Tier: 2 x 1.6 =TB Intel SSD DC P4610 Boot Drive: BOSS controller card + with 2 x M.2 sticks 480 GB |
400-BELT 403-BBTJ |