idfoxdalesystems reference + portfoliolearning in public

project module

Tensor Systems Lab

A cross-language laboratory for matrix and tensor execution across C++, Rust, C, and CUDA kernels.

Problem

Technology lists do not demonstrate how systems behave. This lab will apply the same matrix workload across multiple implementations so memory layout, numerical correctness, execution cost, and portability can be compared directly.

Planned implementations

  • C, C++, and Rust reference kernels with shared numerical fixtures.
  • Native and WebAssembly builds where each execution environment adds useful evidence.
  • CUDA kernels with reproducible benchmark artifacts from a remote GPU runner.
  • A portable JSON format for inputs, outputs, environment metadata, and timing results.
  • A minimal JavaScript adapter only where the browser interface requires it.

Reliability rules

Every implementation must share fixtures, numerical tolerances, and test cases. Benchmark output is evidence only when its environment and Git revision are recorded.