alexziskind1/machine_tests

How alexziskind1/machine_tests works

This repository is a collection of independent performance experiments and benchmark utilities, not one integrated product. The clearest user workflow is to configure and run a benchmark, measure model or sorting performance, and inspect saved results or plots. Other branches cover battery-drain analysis, training examples, and small runtime demonstrations. Source excerpts support only a few cross-module links; unsampled capabilities are represented by their existing directory paths, without inferred pipelines.

Main entry points: config_loader.py MLX Config · main.py LM Studio Models · common.h Sort Algorithms · llm_performance_tester.py Prompt Tester

Components 33

LLM Benchmarks

Prompt Tester ml/auto_prompter/llm_performance_tester.py
Statistical Tester ml/auto_prompter/statistical_llm_tester.py
Prompt Config ml/auto_prompter/config_loader.py
Prompt Results ml/auto_prompter/latest_results.py
Prompt Analysis ml/auto_prompter/analyze_results.py
Prompt Plots ml/auto_prompter/plot_results.py
llama-bench Runner ml/llama_bencher/llama_bench_runner.py
llama-bench Config ml/llama_bencher/config_loader.py
llama-bench Plots ml/llama_bencher/plot_llama_bench_results.py
MLX Runner ml/mlx_bencher/mlx_bench_runner.py
MLX Config ml/mlx_bencher/config_loader.py
MLX Plots ml/mlx_bencher/plot_mlx_bench_results.py
Combined MLX Plots ml/mlx_bencher/plot_combined_bench_results.py
LM Studio Models ml/lmstudio_model_enum_test/main.py

Sorting Benchmarks

Single-thread Sort cpp/sorting_v2/sort_single.cpp
Multi-thread Sort cpp/sorting_v2/sort_multi.cpp
Sort Algorithms cpp/sorting_v2/common.h
Thread Scaling Sort cpp/thread_walker_sort/thread_walker_sort.cpp
Scaling Analysis cpp/thread_walker_sort/analyze.py

Performance Analysis

Battery Data Cleaning battery/analysis/clean_data.py
Battery Drain Plot battery/analysis/plot_battery_drain_over_time.py
Work vs Battery Plot battery/analysis/plot_work_done_vs_battery.py
Efficiency Plot battery/analysis/plot_efficiency.py
Disk Speed Plot diskcopy/plot_speed.py

Training and Demos

Mistral Training ml/mistral_train/lima_train_mistral_fp16.py
Mistral Inference ml/mistral_train/mistral_infer.py
PyTorch Training ml/pytorch/training/train_resnet50_cifar10.py
TensorFlow Training ml/tensorflow/training/train_resnet50_cifar10.py
Python Mandelbrot benchmarksgame/python/mandelbrot/main.py
C# Mandelbrot benchmarksgame/csharp/mandelbrot/MandelbrotCSharp/MandelbrotCSharp/Program.cs
Browser Speedometer browser/speedometer/speedometer.mjs

Other components

Experimenter
LM Studio

Connections 14

Experimenter → Prompt Tester: runs
Experimenter → Statistical Tester: runs
Experimenter → llama-bench Runner: runs
Experimenter → MLX Runner: runs
Experimenter → LM Studio Models: runs ml/lmstudio_model_enum_test/main.py
Prompt Tester → MLX Config: imports ml/auto_prompter/llm_performance_tester.py
Statistical Tester → MLX Config: imports ml/auto_prompter/statistical_llm_tester.py
llama-bench Runner → MLX Config: imports ml/llama_bencher/llama_bench_runner.py
MLX Runner → MLX Config: loads config ml/mlx_bencher/mlx_bench_runner.py
Single-thread Sort → Sort Algorithms: uses cpp/sorting_v2/sort_single.cpp
Multi-thread Sort → Sort Algorithms: uses cpp/sorting_v2/sort_multi.cpp
Thread Scaling Sort → Sort Algorithms: uses cpp/thread_walker_sort/thread_walker_sort.cpp
LM Studio Models → LM Studio: queries models ml/lmstudio_model_enum_test/main.py
LM Studio Models → LM Studio: sends prompts ml/lmstudio_model_enum_test/main.py

Drawn by GitDiagram on November 10, 2026 from the repository's file tree, README and source files. Read as Markdown ·