Hard1 markMultiple Choice

GCP ACE · Question 24 · Domain 3.2: Deploying and implementing GKE resources

You are deploying a GKE Autopilot cluster. You need to deploy a workload that requires specific hardware accelerators (GPUs). Which TWO statements are true regarding this deployment? (Select TWO)

Answer options:

A.

You must manually create a node pool with GPU-enabled machine types.

B.

You must specify the GPU resource requests in your Pod specification.

C.

Autopilot automatically provisions nodes with the requested GPUs.

D.

GKE Autopilot does not support GPUs.

E.

You must install the NVIDIA drivers manually on the Autopilot nodes.

How to approach this question

Understand how GKE Autopilot abstracts infrastructure management compared to Standard mode.

Full Answer

You must specify the GPU resource requests in your Pod specification. Autopilot automatically provisions nodes with the requested GPUs.
GKE Autopilot manages the entire cluster infrastructure. You do not create or manage node pools. To use GPUs, you simply add the appropriate node selector and resource requests (e.g., `nvidia.com/gpu`) to your Pod specification. Autopilot automatically provisions a node with the requested GPU, installs the drivers, and schedules the Pod.

Common mistakes

Applying GKE Standard concepts (manual node pools, manual driver installation) to GKE Autopilot.

Practice the full GCP Associate Cloud Engineer Practice Exam 1

50 questions · hints · full answers · grading

More questions from this exam