vrr/ruvector
2
0
Fork 0
mirror of https://github.com/ruvnet/RuVector.git synced 2026-05-22 19:56:25 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 15
ruvector/crates/profiling
History
Claude c734c0eca5 Reorganize repository structure 
- Move router-* folders into crates/ directory
- Move profiling folder into crates/
- Update Cargo.toml workspace to include new crate locations
- Add node_modules/ and package-lock.json to .gitignore
- Remove node_modules directory from repository
- Create new README.md with project overview and badges
- Move old technical documentation to docs/TECHNICAL_PLAN.md

This reorganization improves the project structure by:
- Consolidating all Rust crates in the crates/ directory
- Following standard Rust workspace conventions
- Cleaning up root directory clutter
- Providing a clear, professional README for new users
2025-11-19 20:53:37 +00:00
..
scripts Reorganize repository structure 2025-11-19 20:53:37 +00:00
README.md Reorganize repository structure 2025-11-19 20:53:37 +00:00

README.md

Ruvector Performance Profiling

This directory contains profiling scripts, reports, and analysis for Ruvector performance optimization.

Directory Structure

profiling/
├── scripts/          # Profiling and benchmarking scripts
├── reports/          # Generated profiling reports
├── flamegraphs/      # CPU flamegraphs
├── memory/           # Memory profiling data
└── benchmarks/       # Benchmark results

Profiling Tools

CPU Profiling

  • perf: Linux performance counters
  • flamegraph: Visualization of CPU hotspots
  • cargo-flamegraph: Integrated Rust profiling

Memory Profiling

  • valgrind: Memory leak detection and profiling
  • heaptrack: Heap memory profiling
  • massif: Heap profiler

Cache Profiling

  • perf-cache: Cache miss analysis
  • cachegrind: Cache simulation

Quick Start

# Install profiling tools
./scripts/install_tools.sh

# Run CPU profiling
./scripts/cpu_profile.sh

# Run memory profiling
./scripts/memory_profile.sh

# Generate flamegraph
./scripts/generate_flamegraph.sh

# Run comprehensive benchmark suite
./scripts/benchmark_all.sh

Performance Targets

  • Throughput: 50,000+ queries per second (QPS)
  • Latency: Sub-millisecond p50 latency (<1ms)
  • Recall: 95% recall at high QPS
  • Memory: Efficient memory usage with minimal allocations in hot paths
  • Scalability: Linear scaling from 1-128 threads

Optimization Areas

  1. CPU Optimization

    • SIMD intrinsics (AVX2/AVX-512)
    • Target-specific compilation
    • Hot path optimization

  2. Memory Optimization

    • Arena allocation
    • Object pooling
    • Zero-copy operations

  3. Cache Optimization

    • Structure-of-Arrays layout
    • Cache-line alignment
    • Prefetching

  4. Concurrency Optimization

    • Lock-free data structures
    • RwLock optimization
    • Rayon tuning

  5. Compile-Time Optimization

    • Profile-Guided Optimization (PGO)
    • Link-Time Optimization (LTO)
    • Target CPU features