Skip to content

Measure useful work.

Infrastructure, runtime, and performance teams

Rubian results separate throughput, IPC, scaling, shell work, responsiveness, and correctness. Project Valux claims require new matched runs.

System work isn’t one metric.

Throughput, latency, scaling, file operations, responsiveness, energy, and cleanup can move in different directions.

Reported results describe component tests.

Check the exact component, version and run record behind a summary. Separate component measurements cannot establish the performance of the complete Valux operating system.

Project Valux performance requires a matched control.

A broader operating-system claim must compare equal work under the same test conditions.

Verify equal work and correct output before comparing performance.

Read next.

Rubian benchmarksResults with the test beside them.

Review public Rubian and Kestowv benchmark summaries, methods, limitations, and the proof required for broader Project Valux claims.

Benchmark methodologyVerify equal work and correct output before comparing performance.

Inspect the public equal-work, correctness, telemetry, repetition, cleanup, and limitation rules used for Rubian and Kestowv testing.

Responsiveness under loadKeep control under load.

Review Rubian and Kestowv responsiveness results, tradeoffs, and the testing required for broader claims.

Working through a hardware or runtime decision? Explore compute evaluations

Rubian/Kestowv Documentation

Startup and discovery, file and project organization, Forge libraries, Taskman, agent work and Kestowv pilot operations.