Optimize SIMD code paths for x86_64 AVX2/AVX-512 and ARM NEON.
npx skills add https://github.com/rust-works/succinctly --skill simd-optimization
Please help me install this Agent Skill. Skill: simd-optimization Source: https://github.com/rust-works/succinctly/tree/main/.claude/skills/simd-optimization Command: npx skills add https://github.com/rust-works/succinctly --skill simd-optimization
Identify and apply SIMD optimization patterns to improve cross-architecture performance on x86_64 and ARM targets.
Provide an initial SIMD optimization pass by identifying a target hot loop and applying architecture-aware vectorization patterns.
High-intent search queries and answers about installing and using this skill.
Apply architecture-aware vectorization patterns to target hot loops, using runtime dispatch strategies and intrinsics to maximize throughput across x86_64 and ARM platforms while maintaining portability.
Use target_feature attributes for architecture-specific tuning and implement runtime feature detection to ensure cross-architecture compatibility while applying AVX2, AVX-512, or NEON intrinsics.
Runtime dispatch strategies detect available CPU features at execution time, allowing your program to select the appropriate SIMD vectorization path, such as AVX-512 or NEON, ensuring optimal performance safely.
Implement SIMD vectorization when facing performance-critical math routines that require measurable speedups, applying architecture-aware patterns to process multiple data points simultaneously for maximum throughput.
Common pitfalls include portability concerns and unsafe usage; apply best practices for target_feature attributes and intrinsics to avoid architecture-specific bugs and ensure cross-architecture compatibility.
mvillmowAnalyze profiling data to identify optimization opportunities in performance-critical code paths.
Generous-CorpEnforces static linting rules and manages CI workflows for macOS x86_64 portability.
bobmatnycOrchestrate coroutines and event loops for non-blocking Python concurrency.
mohitmishra786Analyze auto-vectorization failures and write SSE2, AVX2, and NEON intrinsics.
