# The seven-tool round on the i5-10600KF, 12 Sep 2026

Two phases of harness/field.ps1 (fieldcore.ps1), 10 GiB in 10
files, 768,000-byte slices, 13,990 source blocks, one whole member deleted.

- `field15-phase1.log`: 15% parity, all seven tools, create / verify /
  repair at 1,399 blocks, TWO repetitions, 01:15Z to 04:48Z. Complete.
- `field100-phase2-to-4096.log`: 100% parity, six repairing tools at 1,399 /
  2,098 / 4,096 blocks, ONE repetition, 05:12Z to 11:07Z. Stopped by the maintainer's
  decision after the 4,096 rung to hand the machine to fast-mode work
  (the FIELD-STOPPED-BY-OPERATOR line at the end); par2cmdline and phpar2
  creates skipped by projection (12,000 s against a 1,500 s budget); the
  quick tools' 5,500 to 13,990 depths and a second repetition are DEFERRED.
  A lone parfast leg at 5,500 (53.6 s) is in the log and not on the page.

The first attempt at phase 2 ran unbudgeted from 04:48Z to 05:11Z (see
PARFAST-PUBLICATION-ROUNDS-2026-09-10.md, 12 Sep section) and produced
nothing that is kept.

Page data is generated by the extractor from these two logs
(gate: SHA-256 restoration, never exit code - par2j returns 16).