LLM Task Examples

Provide concise, structured answers for tasks:

• Explain how to compile a C# method to a CUDA kernel and launch it.
• Compare CUDA vs AVX backend tradeoffs for a given workload.
• Diagnose a performance issue given profiler counters.
• Outline data marshalling for a struct array with nested fields.

Format answers with:

• Clear steps and citations to guide/* and reference/* pages
• Definitions from reference/glossary
• Warnings for limitations where relevant