A turnpike booth’s chart requests a GPU for Riva speech containers. A volunteer offers a custom kernel “to make the GPU count.” What is the correct stance?
Select an answer to reveal the explanation.
Short Explanation
GPUs in the chart for official Riva speech are fine; homemade kernels are not the ticket. Use the accelerated service as shipped—do not reinvent the GPU in C++.
Full Explanation
Helm charts may request GPUs for official Riva speech containers. That does not authorize custom CUDA kernel development as a Performance Optimization task. Stay with GPU-backed official services and refuse kernel work.