Compute Shapes

A shape is a template that determines the number of OCPUs, amount of memory, and number of VNICs that are allocated to a compute instance. You choose a shape when you create an instance.

Private Cloud Appliance supports these types of shapes:

  • Standard shapes: Each shape has a fixed number of OCPUs and memory that are allocated to an instance when the instance is created.
  • Flexible shapes: You choose the number of OCPUs and amount of memory that are allocated to an instance.
  • GPU shapes: Each shape has a fixed number of OCPUs, memory, and GPUs that are allocated to an instance when the instance is created.

Standard Shapes

Standard shapes are designed for general-purpose workloads and suitable for a wide range of applications and use cases. Standard shapes provide a balance of cores, memory, and network resources. All standard shapes use block storage for the boot device.

The following table lists the standard shapes:

Shape OCPUs Memory (GB) Maximum VNICs Maximum Bandwidth (Gbps)

VM.PCAStandard1.1

1

16

2

24.6

VM.PCAStandard1.2

2

32

2

24.6

VM.PCAStandard1.4

4

64

4

24.6

VM.PCAStandard1.8

8

128

8

24.6

VM.PCAStandard1.16

16

256

16

24.6

VM.PCAStandard1.24

24

384

24

24.6

VM.PCAStandard1.32

32

512

24

32.0

VM.PCAStandard1.48

48

768

24

48.0

VM.PCAStandard1.60

60

960

24

100.0

Flexible Shapes

A flexible shape lets you customize the number of OCPUs and the amount of memory for an instance. When you update an instance, you can change these properties of the flexible shape, and you can change from a fixed shape to a flexible shape. This flexibility lets you create instances that meet your workload requirements, while optimizing performance and using resources efficiently.

Private Cloud Appliance platform images support the flex shape.

The VM.PCAStandard.E5.Flex shape can be selected only for Private Cloud Appliance X10 systems.

The VM.PCAStandard.E6.Flex shape can be selected only for Private Cloud Appliance X11 systems.

Shape OCPUs Memory (GB) Maximum VNICs Maximum Bandwidth (Gbps)
VM.PCAStandard1.Flex

1–32

64 GB maximum per OCPU

512 GB maximum per instance

1 OCPU: 2 VNICs

2 to 24 OCPUs: 1 VNIC per OCPU

25 to 32 OCPUs: 24 VNICs

1 to 24 OCPUs: 24.6 Gbps

25 to 32 OCPUs: 1 Gbps per OCPU

VM.PCAStandard.E5.Flex

1–96

64 GB maximum per OCPU

960 GB maximum per instance

1 OCPU: 2 VNICs

2 to 24 OCPUs: 1 VNIC per OCPU

25 to 96 OCPUs: 24 VNICs

1 to 24 OCPUs: 24.6 Gbps

25 to 40 OCPUs: 1 Gbps per OCPU

41 to 96 OCPUs: 40.0 Gbps

VM.PCAStandard.E6.Flex

1–96

64 GB maximum per OCPU

960 GB maximum per instance

1 OCPU: 2 VNICs

2 to 24 OCPUs: 1 VNIC per OCPU

25 to 96 OCPUs: 24 VNICs

1 to 24 OCPUs: 24.6 Gbps

25 to 40 OCPUs: 1 Gbps per OCPU

41 to 96 OCPUs: 40.0 Gbps

GPU Shapes

GPU shapes are designed for enterprise GPU-accelerated workloads, such as AI and graphics intensive applications. These shapes are available on systems with an optional GPU expansion rack, when an Oracle Linux 8 or Oracle Linux 9 platform image is selected to create the instance. Instances created with a GPU shape have direct (passthrough) access to 1-4 physical GPUs. The ratio between GPUs, OCPUs and memory is fixed.

The following table lists the GPU shapes:

Shape GPUs OCPUs Memory (GB) Maximum VNICs Maximum Bandwidth (Gbps)

VM.GPU.L40S.1

1

27

VM: 200

GPU: 48

24

400

VM.GPU.L40S.2

2

54

VM: 400

GPU: 96

24

400

VM.GPU.L40S.3

3

81

VM: 600

GPU: 144

24

400

VM.GPU.L40S.4

4

108

VM: 800

GPU: 192

24

400