OCI Data Science now offers flex shapes across resources

August 25, 2022 | 4 minute read
Wendy Yip
Data Scientist
Text Size 100%:

In July and August 2022, Oracle Cloud Infrastructure (OCI) Data Science launched support for new Compute shapes across several resources in the Data Science service. Notebook sessions, jobs, and model deployments within Data Science now support the following flexible CPU shapes:

Name

Instance name

Cores (OCPU)

Memory (GB)

AMD E3

VM.Standard.E3.Flex

1-64

1-1024

AMD E4

VM.Standard.E4.Flex

1-64

1-1024

Intel X9 Optimized

VM.Optimized3.Flex

1-18

1-256

Intel X9

VM.Standard3.Flex

1-32

1-512

What is a flexible shape?

A flexible shape is a shape that lets you customize the number of OCPUs and the amount of memory when launching or resizing your virtual machine (VM). When you create the VM instance, you can select the number of OCPUs and amount of memory to run your workload.

Data Science notebooks, jobs, and model deployments now support E3 and E4 flex shapes. E3 flex shapes are based on the second-generation AMD EPYC processors, while E4 flex shapes are based on the third-generation AMD EPYC processors. Performance benchmarks show that E4 shapes deliver an estimated 15% increase in integer performance and an estimated 21% increase in floating point performance compared to E3 instances. For both shapes, you can choose from 1–64 OCPUs and up to 1 TB in memory.

We’re also introducing two flexible shapes based on Intel’s 3rd Generation Xeon Ice Lake processors. Intel X9 Optimized is the first shape based on the Ice Lake processor supported by OCI Compute and is designed for compute-intensive and high-performance computing workloads. You can choose from 1–18 OCPUs and up to 256 GB in memory. The Intel X9 or Standard3 instance is the latest addition from the Ice Lake processor family and an improvement over the Standard2 instance we already offer. Standard3 instance provides up to 60% better performance than Standard 2 instance for the same price. You can choose from 1–32 OCPUs and up to 512 GB in memory.

How to use flex shapes

When you log into OCI Data Science, you see an updated UI that allows you to pick the new supported shape. The following images from the “Select compute” panel during the notebook creation workflow show the availability of the new shapes.

The following graphics show the shape offerings with the Intel Ice Lake processors in the “Select compute” panel during the Notebook session creation.

A screenshot of the Select Compute screen showing the shape offerings with the Intel Ice Lake processors.

We offer the following shapes with the AMD EPYC processors in the "Select compute" panel.

A screenshot of the Select Compute panel with the the E4 Flex shape selected.

A screenshot of the Select Compute panel with the E3 Flex shape selected.

Conclusion

With the introduction of the new Compute shapes, we provide greater flexibility and better price performance for our customers.

For more information, see the following resources:

Wendy Yip

Data Scientist


Previous Post

Using Jupyter on OCI Data Science to analyze HPC results

Joe Hahn | 7 min read

Next Post


Set up autoscaling for OCI Data Science model deployment

Tzvi Keisar | 15 min read