Coordinate profiling delegation and generate structured performance analysis reports.
npx skills add https://github.com/NVIDIA/skills --skill perf-analysis-nvidia
Please help me install this Agent Skill. Skill: perf-analysis Source: https://github.com/NVIDIA/skills/tree/main/skills/TensorRT-LLM/perf-analysis Command: npx skills add https://github.com/NVIDIA/skills --skill perf-analysis-nvidia
Performance analysis coordination eliminates the need for every team to implement bespoke profiling workflows by guiding the delegation to specialists and standardizing outputs.
Delegate profiling tasks to the perf-profiling-specialist and generate a structured performance analysis report.
High-intent search queries and answers about installing and using this skill.
To diagnose GPU performance bottlenecks, this Skill coordinates profiling tasks to classify compute, memory, launch, communication, or sync issues, then generates a structured report with actionable optimization recommendations.
You can generate a structured performance profiling report by delegating workload analysis tasks to domain experts, which then produces standardized outputs containing Summary, Metrics, Findings, and Recommendations sections.
Performance analysis coordination orchestrates workload profiling by delegating tasks to specialized tools, standardizing outputs to eliminate the need for bespoke workflows when analyzing compute, memory, and interconnect dimensions.
Yes, you can profile compute, memory, and interconnect dimensions simultaneously by delegating validated profiling tasks to specialists, ensuring comprehensive bottleneck classification across all dimensions in the final report.
The best way to standardize performance metrics reporting is by orchestrating profiling delegation to specialists, ensuring all validated metrics are compiled into a standardized Summary, Metrics, Findings, and Recommendations format.
Yes, you need specialized profiling tools to identify communication and sync bottlenecks, as this Skill coordinates the safe delegation of profiling tasks to those domain experts to extract validated metrics.
dotruruConfigure Zig builds for SIMD, cross-compilation, and CUDA interop.
Wenwen555Profile CPU, memory, GPU, and interconnect bottlenecks and generate structured performance reports.
jandan138Profile Python scripts, processes, GPU workloads, and interconnect topologies into structured reports.
fableindigo-gifDetects bottlenecks in Claude Flow swarms and generates optimization recommendations.
