A systems administrator is troubleshooting a VDI deployment that is used to run high-frame-rate rendering. Users are reporting frequent application crashes. After running a benchmark, the administrato

GPU compute mode (CUDA-only or compute-exclusive mode) is designed for scientific computing and AI/ML workloads, not for graphical rendering pipelines, and would likely worsen rendering application crashes.

Allocating more RAM addresses CPU-side memory pressure, not GPU resource limitations - the benchmark context points to GPU resource exhaustion as the cause of the crashes.

In a VDI environment, vGPU profiles define the amount of GPU frame buffer memory and compute capacity assigned to each virtual machine. High-frame-rate rendering is GPU-intensive and requires substantial VRAM - an undersized profile causes the GPU to be oversubscribed, leading to application instability and crashes. Selecting a higher vGPU profile increases per-VM resource allocation, directly addressing the root cause identified by the benchmark.

Graphics mode is already the appropriate GPU mode for rendering workloads - the issue is insufficient resource allocation within that mode, not the mode configuration itself.