Skill UI

This guide outlines the necessary steps to ensure that parallel computing performance papers are scientifically reproducible. Since parallel results depend heavily on machine architecture, topology, and runtime conditions, proper documentation must cover everything from CPU/GPU specifications and interconnects to thread pinning policies, random seeds, and compiler flags. The focus is on proving the scaling trend, not just the absolute numbers, allowing results to hold true across different hardware platforms.

This comprehensive guide outlines the strict criteria for submitting a paper to the SenSys conference after the 2026 merger. It clarifies that the core contribution must be a built, measured sensing or embedded system, demonstrating value based on physical resources like energy, memory, and latency on real hardware, rather than being a pure algorithm or networking mechanism.

A comprehensive guide for designing and auditing high-stakes networking experiments, particularly for top-tier academic submissions like SIGCOMM. It outlines the 'evidence ladder' from microbenchmarks to live deployment, emphasizing the need to select robust baselines, report full performance distributions (percentiles and variance), and accurately map the mechanism's operational break points.

This comprehensive guide outlines best practices for designing data systems evaluations, ensuring results are robust, reproducible, and convincing to expert reviewers. It covers workload realism, baseline fairness, necessary ablation studies, and advanced metric visualization techniques (e.g., sweeping curves, failure analysis) required for top-tier data systems research.

This guide outlines the best practices for preparing system artifacts for academic evaluation, following the ACM Artifact Review and Badging scheme. It details how authors must ensure reproducibility, especially for tail latency and cost results, by providing scaled-down testbeds, comprehensive documentation, and clear provenance tracking to satisfy conference evaluators.

This comprehensive guide outlines best practices for achieving high levels of scientific reproducibility in systems papers, especially for cloud measurement studies. It teaches authors how to map every claim in their paper to verifiable evidence, detailing the entire testbed setup, reproducing tail latency and costs accurately, and maintaining consistency between reported numbers and artifacts.

A comprehensive guide for authors submitting to high-impact scientific journals. It outlines the essential principles for structuring experiments and presenting evidence, emphasizing claims-evidence matching, fair benchmarking, differentiating result sources (simulation, hardware, theory), and ensuring statistical rigor. Covers best practices across computer science, control, communication, and microelectronics.

A comprehensive guide detailing how to package and present research artifacts—including code, hardware designs, and datasets—for major academic conferences. It outlines best practices for both initial review-time evidence and long-term public reuse, ensuring reproducibility and maximizing the chances of adoption by future builders.