Noiseblind

Detailed_Analysis

Senior Capstone // Instruction-Level Obfuscation

Execution_Environment

C++20 // AVX-2 SIMD // Libsodium // FastNoise2 // CLI11 // Dear ImGui // CMake // nlohmann/json

Investigation_Report

Noiseblind layers deterministic noise from a password-derived key. Initial development revealed that the obfuscation was "annoyingly slow," prompting an investigation into micro-architectural optimizations.

After encountering integration issues with FastNoise2, I developed MercurySIMD. This custom wrapper resolved the execution stalls and forced a shift in focus to the next bottleneck: Hardware I/O.

I subsequently engineered Polymorphic Concurrency Pipelines. The system now detects drive physics—deploying barrier-based producer/consumer for rotational media (HDD) or work-stealing for flash-based storage (SSD).

Optimization_Log

// Initialization

Senior capstone project: multi-layered deterministic noise generation via Argon2. Identified critical throughput bottlenecks during early-stage obfuscation testing.

// Throughput_Scaling

Engineered MercurySIMD, a custom AVX2 wrapper. Achieved a 9x performance increase, shifting the system bottleneck from CPU cycles to HDD I/O latency.

Primary_Document_Stream