MojoCPUs: Experimental Data and Benchmark Code for "Memory Layout Matters: An Empirical Study of How Memory Organization Shapes the Performance–Safety–Productivity Trade-off"

This repository contains the complete experimental data and benchmark code accompanying the paper "Memory Layout Matters: An Empirical Study of How Memory Organization Shapes the Performance–Safety–Productivity Trade-off." The dataset includes 29 controlled experiments spanning six programming environments (C, Rust, Mojo, Zig, Julia, Python) across five DLAM evaluation dimensions: Performance, Safety, Code Complexity, Development Cost, and Ecosystem. Experiments cover micro-benchmarks (matmul, prefix sum, dot product, softmax, N-body simulation), compiler ablation studies (GCC vs. Clang/LLVM), energy efficiency profiling, cross-language single-thread performance comparisons, progressive optimization from L0 to L4, memory layout analysis (flat vs. nested), metaprogramming efficiency, parallel computation benchmarks, software evolution cost measurement, compilation iteration efficiency analysis, and cross-scale generalization validation. All benchmark source code (C, Rust, Mojo, Zig, Julia, Python), measurement scripts, and result logs are included.