A team trains a model with 10 billion parameters on 64 GPUs using data parallelism. The GPUs are connected via 25 Gbps Ethernet. Profiling shows that gradient synchronization takes 70% of the training time. The team wants to reduce communication overhead without changing the model or the number of GPUs. Which technique would most directly reduce the communication time?

Enable gradient compression (e.g., top-k sparsification or quantization) to reduce the data sent during synchronization.

Switch to model parallelism to reduce the number of gradients.

Increase the batch size to reduce the frequency of synchronization.

Use a single GPU to avoid communication.

NetAppCertified AI Expert

Domain 5Objective 6

Describe Solutions to Maximize Performance in AI Workloads CERTIFIED-AI-EXPERT Practice Questions (Page 8)

Part of the AI Common Challenges domain, which accounts for 22% of the CERTIFIED-AI-EXPERT exam. NetApp does not publish an official question count, but from its 90-minute exam (~35–60 total, ~8–13 in this domain), expect 1–2 from this objective — we provide 36 practice questions to prepare you well beyond it. (estimate)

36questions here

8free pages

10concepts

22%of the exam

10 Concepts in This Objective

Performance Bottleneck Identification
Data Pipeline Optimization
Compute Resource Scaling
GPU Utilization Maximization
Storage Performance Tuning
Network Optimization for Distributed Training
Model Parallelism and Data Parallelism
Caching and Prefetching Strategies
Performance Monitoring and Profiling
Trade-offs in Performance Optimization

Questions 36–36

36expert · hard
A team trains a model with 10 billion parameters on 64 GPUs using data parallelism. The GPUs are connected via 25 Gbps Ethernet. Profiling shows that gradient synchronization takes 70% of the training time. The team wants to reduce communication overhead without changing the model or the number of GPUs. Which technique would most directly reduce the communication time?
Select an answer first

Finished these 1 questions?

Review the revealed explanations, or continue through the curriculum.

No more pagesBack to CERTIFIED-AI-EXPERT

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by NetApp. “CERTIFIED-AI-EXPERT” is a trademark of its owner, used for identification only.

Describe Solutions to Maximize Performance in AI Workloads CERTIFIED-AI-EXPERT Practice Questions (Page 8)

Questions 36–36

Keep studying — AI Common Challenges