II.
LibrarySkill overview
Reference · livelib-skill:gpu-programming--cuda-graphs
cuda-graphs overview
Expert skill for CUDA Graph capture and optimization for reduced launch overhead. Capture CUDA operations into graphs, instantiate and execute graph instances, update graph node parameters, profile graph vs stream execution, design graph-friendly kernel patterns, and optimize launch latency for inference.
Attributes
displayName
cuda-graphs
description
Expert skill for CUDA Graph capture and optimization for reduced launch overhead. Capture CUDA operations into graphs, instantiate and execute graph instances, update graph node parameters, profile graph vs stream execution, design graph-friendly kernel patterns, and optimize launch latency for inference.
libraryPath
library/specializations/gpu-programming/skills/cuda-graphs/SKILL.md
specialization
gpu-programming
contentSummary
# cuda-graphs
You are **cuda-graphs** - a specialized skill for CUDA Graph capture and optimization. This skill provides expert capabilities for reducing kernel launch overhead and optimizing execution patterns through graph-based workflows.
## Overview
This skill enables AI-powered CUDA Graph op
Outgoing edges
lib_applies_to_domain1
- domain:scientific-computing·DomainScientific Computing
lib_belongs_to_specialization1
- specialization:gpu-programming·Specialization
lib_involves_role2
- role:computational-scientist·RoleComputational Scientist
- role:ml-engineer·RoleMachine Learning Engineer
lib_requires_skill_area3
- skill-area:cuda-kernels·SkillAreaCUDA Kernel Programming
- skill-area:compute-shaders·SkillAreaCompute Shaders
- skill-area:profiling-cuda·SkillAreaCUDA Profiling
Incoming edges
None.