Skip to content

Tools landscape

Reviewed 3 October 2026. This page lists the tools that practitioners and researchers use and cite for network-time and clock analysis. For each, it says what the tool is good at, where ntpstats overlaps, and how the two work together. ntpstats aims to interoperate with these tools, not to replace them. Corrections are welcome as issues or pull requests.

Frequency-stability analysis (metrology)

Tool What it is Relation to ntpstats
Stable32 (W. Riley; free from IEEE UFFC since 2017) The reference Windows program for stability analysis, cited throughout the literature Data files read and written (-f stable32-phase, convert --to stable32-phase), so a dataset can be analysed in both; the NIST SP 1065 tables it is checked against are reproduced to 7 digits, and Stable32's own output files on four data sets agree to its printed digits (Validation)
TimeLab (J. Miles, KE5FX) Acquisition and plotting for counters and phase-noise analysers; the time-nuts favourite .tim import waits for sample files, as the format is not documented (#42); HTOT and the other TimeLab statistics are available
allantools (Python, LGPL) Library of ADEV-family statistics; used in many papers and in BIPM training Independent implementation, checked against a frozen allantools table; from ntpstats.compat import allantools offers its function signatures (migrating)
MATLAB allanvar, R allanvar Generic Allan variance functions ntpstats adds EDF-based CIs, noise identification, gap handling and time-transfer inputs

NTP / NTS operations

Tool What it is Relation to ntpstats
chrony / chronyc The default client on most Linux distributions; logs tracking, measurements, statistics and refclocks All of these logs are parsed, and chrony can be polled live
ntpd-rs Rust NTP/NTS daemon with experimental NTPv5 and, in 2.0, CSPTP; announced as the default in Ubuntu 27.04 NTPv5 interop tested; polled live with ntpstats watch ntpd-rs (1.x metrics) or read from Prometheus with ntpstats prom; its CSPTP exchanges are read from captures (Sources, PTP)
NTPsec ntpviz Percentile plots from loopstats/peerstats Same inputs; ntpstats adds stability statistics, network metrics and chrony support
ntpxyz (Python) Plots of ntpd loopstats, sysstats and usestats with matplotlib Overlapping inputs for loopstats; ntpstats adds confidence intervals, peer/network analysis and the other daemons
chrony_ntp_logconv Converts chrony tracking/statistics logs so that ntpviz can plot them Not needed with ntpstats, which reads chrony logs directly (and keeps chrony's error bounds distinct from an Allan deviation)
ntpperf (M. Lichvar) Load and timestamp-accuracy tester for NTP servers and PTP masters Complementary: ntpstats analyses the resulting NTP and PTP captures
chrony_exporter + Prometheus/Grafana Operational dashboards ntpstats exports OpenMetrics and OpenTelemetry with a Grafana dashboard (Monitoring) and imports Prometheus range queries (ntpstats prom)

PTP and telecom

Tool What it is Relation to ntpstats
linuxptp (ptp4l, phc2sys, ts2phc, pmc) The Linux PTP stack Logs parsed; live polling via pmc (ntpstats watch ptp4l)
Meinberg PTP Track Hound (free, closed source) Captures and decodes PTP traffic, groups devices, monitors Complementary: ntpstats computes offset, PDV and time-error statistics from the captures (PTP & time error)
Calnex CAT, VIAVI, Keysight, Microchip software Vendor analysis for test equipment: TE, cTE, dTE, MTIE, TDEV, FPP with standard masks ntpstats offers transparent, scriptable TE/cTE/dTE/MTIE/TDEV on exported data (PTP & time error); exports are read through TOML import profiles, with vendor profiles added from contributed samples (#35)
facebook/time (ptpcheck, sptp, ptp4u, fbclock) Meta's open-source PTP/NTP tools and TrueTime-style uncertainty ptpcheck stats live polling (ntpstats watch ptpcheck); fbclock windows validated with ntpstats bounds; an SPTP-style client model is scored in the bench (Trace replay & PTP)
PTP-DAL (Python) Offline analysis of recorded PTP datasets: synchronisation algorithms scored with max|TE| and MTIE Similar research purpose; ntpstats works from standard captures and daemon logs, simulates exchanges with ground truth and covers NTP too
timeTools (Python, GPL-3) Time error, MTIE and TDEV, and PDV generation with ITU-T G.8263-style methods Overlapping metrics; ntpstats (MIT) adds the input formats, limits and masks, captures and the audit report
OCP TAP Time Card / Open Time Server Open-hardware grandmaster (GNSS + atomic oscillator) Holdover prediction applies to its logs (Metrology); dedicated parsers wait for sample data (#21)
AWS ClockBound Clock-error bound daemon for EC2 Bounds validated against a reference (ntpstats bounds)
Windows w32tm The Windows time service w32tm /stripchart output read directly
White Rabbit (CERN OHWR) Sub-ns time transfer Its phase logs can be analysed as phase data; a dedicated format can be added as a plugin (Writing a plugin)
Wireshark General capture and dissection ntpstats reads the same pcap/pcapng files and computes the timing statistics

GNSS and time transfer

Tool What it is Relation to ntpstats
CGGTTS tools (BIPM r2cggtts, OpenTTP, rtk-rs cggtts) GNSS common-view time transfer between laboratories CGGTTS files read with checksums verified; common-view and all-in-view differencing (ntpstats cv) (Research data)
RTKLIB, IGS clock products, BIPM Circular T GNSS processing; satellite/station clocks; UTC−UTC(k) RINEX clock files and Circular T read directly (Research data)

Research platforms and datasets

Tool What it is Relation to ntpstats
OMNeT++/INET (gPTP models), ns-3 Network simulators used for TSN/PTP studies .vec result files read (INET clocks become time error) and written; INET oscillator settings fitted to a real clock; ns-3 text and captures (Trace replay & PTP chains)
RIPE Atlas NTP measurements, NTP Pool monitor data Internet-scale public measurements used in recent studies (PACMCS 2024, NDSS 2026) Both read directly (Research data); ntpstats publishes its own monthly interop measurements as an open dataset (data/interop/)

Where ntpstats is different

  • One tool across the chain: daemon logs, packet captures, instruments and live measurement (NTPv4, NTS, NTPv5) feed the same statistics, confidence intervals and reports.
  • Validation, not only plotting: a simulator with ground truth, reference estimators and benchmarks, so an algorithm or a device can be scored.
  • Open and dependency-light: MIT-licensed, numpy only, runs anywhere Python runs (and in the browser), a small web UI, and scriptable in CI with a stable Python API.