Pipeline Gpu Kernel
Skill Verified ActiveApply software pipelining (double-buffering) to a tiled GPU kernel to overlap global memory loads with Tensor Core computation. Covers prologue/loop/epilogue restructuring, LDG-register vs cp.async (LDGSTS) variant selection based on compute/load ratio, shared memory budget verification against architecture-specific occupancy cliffs, and SASS-level verification of load/compute overlap.
Optimize GPU kernel performance by implementing advanced software pipelining techniques to effectively overlap memory operations with computation.
Features
- Software pipelining for GPU kernels
- Double-buffering of shared memory
- Variant selection based on compute/load ratio
- Analysis of load/compute overlap in SASS
- Shared memory budget verification against occupancy cliffs
Use Cases
- When a GPU kernel is identified as memory-bound.
- When warp interleaving alone is insufficient to hide DRAM latency.
- When restructuring a sequential load-sync-compute-sync kernel loop.
- When needing to optimize Tensor Core computation by overlapping memory loads.
Non-Goals
- Optimizing kernels that are not memory-bound.
- Addressing bottlenecks unrelated to memory loads or Tensor Core computation.
- Applying pipelining to kernels without a distinct load and compute phase.
- Basic CUDA compilation; assumes familiarity with `nvcc` and GPU architectures.
Practical Utility
- info:Usage examplesWhile the SKILL.md provides detailed procedural steps, it lacks concrete end-to-end invocation examples with specific inputs and expected outputs for the CUDA kernel optimization.
Installation
/plugin install agent-almanac@pjt222-agent-almanacQuality Score
VerifiedTrust Signals
Similar Extensions
Performance Analysis
100Comprehensive performance analysis, bottleneck detection, and optimization recommendations for Claude Flow swarms
MongoDB Connection Optimizer
100Optimize MongoDB client connection configuration (pools, timeouts, patterns) for any supported driver language. Use this skill when working/updating/reviewing on functions that instantiate or configure a MongoDB client (eg, when calling `connect()`), configuring connection pools, troubleshooting connection errors (ECONNREFUSED, timeouts, pool exhaustion), optimizing performance issues related to connections. This includes scenarios like building serverless functions with MongoDB, creating API endpoints that use MongoDB, optimizing high-traffic MongoDB applications, creating long-running tasks and concurrency, or debugging connection-related failures.
Sql Optimization
100Universal SQL performance optimization assistant for comprehensive query tuning, indexing strategies, and database performance analysis across all SQL databases (MySQL, PostgreSQL, SQL Server, Oracle). Provides execution plan analysis, pagination optimization, batch operations, and performance monitoring guidance.
Core Web Vitals
100Optimize Core Web Vitals (LCP, INP, CLS) for better page experience and search ranking. Use when asked to "improve Core Web Vitals", "fix LCP", "reduce CLS", "optimize INP", "page experience optimization", or "fix layout shifts".
Analyze Kernel Bottleneck
99Systematically identify whether a GPU kernel is compute-bound, memory-bound, or latency-bound using roofline analysis, occupancy calculations, compute/load ratio per tile, and SASS instruction inspection. Produces a decision matrix for optimization strategy selection (cp.async, warp interleaving, tiling, double-buffering, or CuAssembler hand-tuning).
Vector Index Tuning
99Optimize vector index performance for latency, recall, and memory. Use when tuning HNSW parameters, selecting quantization strategies, or scaling vector search infrastructure.