Uploaded image for project: 'OCMUI - OpenShift Cluster Manager UI'
  1. OCMUI - OpenShift Cluster Manager UI
  2. OCMUI-1627

[OSD-GCP] UI Epic for - Enable NVIDIA A100-enabled A2 machine type

    • False
    • False
    • 1
    • OCM Core Sprint 252, OCM Core Sprint 253, OCMUI Core Sprint 254

      UI changes and Scope:

      <this is based on OCM backend scope, needs to be refined for UI scope needed for this feature>

      This ticket is for enabling NVIDIA A100-enabled A2 machine type. For more details refer XCMSTRAT-175. 

      • Scope of this ticket: CCS only (both Annual and GCP Marketplace)
      • Customers should be able to create clusters with GPU instances as well as add new GPU instances
      • The quota rules for Annual and GCP Marketplace will be the same. 
      • Externally, no new SKUs should be required. The customers should be able to use existing SKUs/entitlements to use both standard non-GPU and GPU instances with the CCS billing model

      Not considering a2-ultragpu-8g as this is not currently listed as supported instance type in OCP 4.14, 4.15 docs

      https://docs.openshift.com/container-platform/4.14/machine_management/creating_machinesets/creating-machineset-gcp.html#machineset-gcp-enabling-gpu-support_creating-machineset-gcp

       

      https://cloud.google.com/compute/docs/gpus#a100-gpus

      Instance-id generic name name CPU memory Closest AWS EC2 instance with
      similar GPU-vCPU-memory
      a2-highgpu-1g a2-highgpu-1g a2-highgpu-1g-nvidia-a100 - Accelerated Computing 12 85 p4d.24xlarge
      (8 GPUs, 96 vCPUs, 1152 GBi memory)
      a2-highgpu-2g a2-highgpu-2g a2-highgpu-2g-nvidia-a100 - Accelerated Computing 24 170 p4d.24xlarge
      (8 GPUs, 96 vCPUs, 1152 GBi memory)
      a2-highgpu-4g a2-highgpu-4g a2-highgpu-4g-nvidia-a100 - Accelerated Computing 48 340 p4d.24xlarge
      (8 GPUs, 96 vCPUs, 1152 GBi memory)
      a2-highgpu-8g a2-highgpu-8g a2-highgpu-8g-nvidia-a100 - Accelerated Computing 96 680 p4d.24xlarge
      (8 GPUs, 96 vCPUs, 1152 GBi memory)
      a2-megagpu-16g a2-megagpu-16g a2-megagpu-16g-nvidia-a100 - Accelerated Computing 96 1360 p4d.24xlarge
      (8 GPUs, 96 vCPUs, 1152 GBi memory)
      a2-ultragpu-8g a2-ultragpu-8g a2-ultragpu-8g-nvidia-a100 - Accelerated Computing 96 1360 p4d.24xlarge
      (8 GPUs, 96 vCPUs, 1152 GBi memory)

              emingora Enrique Mingorance Cano
              rh-ee-smulkutk Shreyans Mulkutkar
              Akash Kanni
              Aleš Pecha Aleš Pecha
              Votes:
              0 Vote for this issue
              Watchers:
              8 Start watching this issue

                Created:
                Updated:
                Resolved: