nerdexam
CompTIA

CV0-002 · Question #469

A cloud engineer is upgrading a high-performance computing cluster for the private cloud. The existing cluster is being replaced with GPU servers. A single GPU server is capable of the same teraflops

The correct answer is D. 9 GPU servers, 10 CPU servers. The cloud engineer needs to select a new GPU/CPU server configuration that provides a total of 100 teraflops, matching the baseline performance of the existing cluster.

Cloud architecture

Question

A cloud engineer is upgrading a high-performance computing cluster for the private cloud. The existing cluster is being replaced with GPU servers. A single GPU server is capable of the same teraflops output as 10 CPU servers. The current cluster configuration is as follows:

  • 100 quad-core CPU servers capable of producing 100 teraflops.
  • The baseline and current usage is 100%.

The new cluster was set up and benchmarked in four different configurations. Which of the following configurations will meet the baseline teraflops performance of the cluster while maintaining the current usage?

Options

  • A1 GPU server, 80 CPU servers
  • B2 GPU servers, 50 CPU servers
  • C5 GPU servers, 40 CPU servers
  • D9 GPU servers, 10 CPU servers

How the community answered

(33 responses)
  • A
    12% (4)
  • B
    30% (10)
  • C
    6% (2)
  • D
    52% (17)

Why each option

The cloud engineer needs to select a new GPU/CPU server configuration that provides a total of 100 teraflops, matching the baseline performance of the existing cluster.

A1 GPU server, 80 CPU servers

This configuration provides 1 GPU server (10 teraflops) + 80 CPU servers (80 teraflops) = 90 teraflops, which is less than the required 100 teraflops.

B2 GPU servers, 50 CPU servers

This configuration provides 2 GPU servers (20 teraflops) + 50 CPU servers (50 teraflops) = 70 teraflops, which is significantly less than the required 100 teraflops.

C5 GPU servers, 40 CPU servers

This configuration provides 5 GPU servers (50 teraflops) + 40 CPU servers (40 teraflops) = 90 teraflops, which is less than the required 100 teraflops.

D9 GPU servers, 10 CPU serversCorrect

This configuration consists of 9 GPU servers and 10 CPU servers. Since each GPU server produces 10 teraflops and each CPU server produces 1 teraflop, this totals (9 * 10 teraflops) + (10 * 1 teraflop) = 90 + 10 = 100 teraflops, which exactly meets the required baseline performance.

Concept tested: Resource capacity calculation and server equivalency

Topics

#HPC#GPU computing#Capacity planning#Resource optimization

Community Discussion

No community discussion yet for this question.

Full CV0-002 Practice