NCA-AIIO Question 115
Select 3An organization plans to virtualize its accelerated infrastructure to support AI workloads. Which considerations are critical to ensure optimal performance and resource utilization?
- A
Ensuring GPU passthrough or vGPU support is configured correctly for virtual machines.
- B
Prioritizing virtual machines with higher CPU resources while ignoring GPU allocation.
- C
Evaluating hypervisor compatibility with NVIDIA GPU drivers and CUDA versions.
- D
Configuring memory oversubscription to maximize virtual machine density.
- E
Enabling appropriate PCIe bandwidth allocation to prevent GPU bottlenecks.
Show answer and explanation
Correct answers: A, C, E
Explanation
Virtualizing accelerated infrastructure for AI workloads requires careful consideration of GPU-specific configurations, such as passthrough or vGPU support, hypervisor compatibility with GPU drivers, and PCIe bandwidth allocation. These factors ensure that the virtualized environment can fully utilize the GPU's power without introducing performance bottlenecks. Neglecting these aspects can lead to degraded AI workload performance or system inefficiencies.
- A. Correct.
Ensuring GPU passthrough or vGPU support is critical as it allows virtual machines to directly access GPU hardware, essential for AI and accelerated workloads.
- B. Incorrect.
While CPU resources are important, ignoring GPU allocation would lead to suboptimal AI workload performance, as GPUs are central to these tasks.
- C. Correct.
Hypervisor compatibility with NVIDIA GPU drivers and CUDA versions ensures that the virtualized environment can properly utilize the GPUs for accelerated workloads.
- D. Incorrect.
Memory oversubscription may increase virtual machine density but can lead to memory contention issues, which can degrade GPU-accelerated performance.
- E. Correct.
PCIe bandwidth allocation is vital to prevent data transfer bottlenecks between the GPU and the rest of the system, directly impacting AI workload performance.