Stress Test the GPU Nodes POWEREDGE-XE-OPERATE Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   After a successful GPU stress test, you want to ensure that the GPU is stable for production use. Which additional check should you perform?

   • ARun a short workload and verify that the GPU performs as expected.
   • BRun `nvidia-smi` to confirm the GPU is still visible.
   • CCheck the GPU temperature after the test to ensure it returns to idle levels.
   • DReview the stress test logs for any warnings.
   Reveal Solution Discussion Select an answer first
2. 12expert · hard

   During a stress test, you notice that GPU temperature is rising steadily and has reached 95°C, close to the maximum. The test is still running. What should you do?

   • AReduce the stress test intensity to lower the temperature.
   • BStop the stress test immediately to prevent potential hardware damage.
   • CIgnore the temperature and focus on the error count.
   • DLet the test continue to see if the GPU can handle the heat.
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A data scientist reports intermittent GPU errors on a PowerEdge XE server running a training workload. Before running a stress test, you want to confirm that the system is healthy and that the GPU is visible to the OS. Which command sequence should you use first?

   • ARun `furmark` to stress the GPU and check for visual artifacts.
   • BRun `nvidia-smi dmon` to monitor real-time metrics, then start the stress test.
   • CRun `gpu-burn` immediately to load the GPU and observe if errors appear during the test.
   • DRun `nvidia-smi` to check GPU visibility and driver version, then run `dmesg | grep -i nvidia` to check for driver errors.
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   When running a GPU stress test, what is the purpose of specifying a duration parameter (e.g., `gpu-burn 60`)?

   • ATo control how long the stress test runs
   • BTo limit the test to a specific GPU
   • CTo set the target temperature for the test
   • DTo set the number of GPU cores used
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   You need to stress test a PowerEdge XE server with four NVIDIA GPUs to validate thermal performance under maximum load. The test should run for 30 minutes and log metrics for later analysis. Which approach should you use?

   • ARun `furmark` with a 30-minute timer and rely on its built-in logging.
   • BRun `nvidia-smi --loop=1` for 30 minutes to monitor metrics while the GPU idles.
   • CRun `gpu-burn 600` and then check the output for errors after completion.
   • DRun `gpu-burn 1800` and simultaneously run `nvidia-smi dmon -s pucvmet -o DT` to capture metrics.
   Reveal Solution Discussion Select an answer first