• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

super3 / llmjob / 44350093850
100%

Build:
DEFAULT BRANCH: main
Ran 18 Nov 2026 04:42PM UTC
Jobs 1
Files 28
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

18 Nov 2026 04:41PM UTC coverage: 100.0%. Remained the same
44350093850

push

github

web-flow
Release v0.7.5: RX 9000 and RTX 50 speedups, HeroMiners old-pool balance (#307)

* RDNA 4: the 16 GB RX 9060 XT keeps the old tile order at batch 128

Since 7c283621 the 16 GB 9060 XT mines k 8192 at a column batch of 128, so its grid is 8 column tiles wide. The 4 x 8
supertile walk was tuned at a batch of 256. At 128 the old order (any walk 8 tiles wide) read +0.4% and +0.45% in
hashrate.js on two PCs (ac3f08f2, 8aca3ffa), ahead in every round, twins within 0.1%. The walk now stays off on
gfx1200 when the grid is under 16 column tiles, so the 8 GB card's batch of 256 still walks.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018A6hR4v2YEpsJuAoZzfi3r

* benchmark.md: the 16 GB RX 9060 XT's old tile order at batch 128

Records the three runs behind 66843e81 (+0.4%, +0.45% and +0.8% on two
PCs, ahead in every round) and its 400 of 400 verify-hits.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018A6hR4v2YEpsJuAoZzfi3r

* RDNA 4: time k 8192 against k 16384 on each RX 9070 XT PC and keep the faster

k 16384 read +0.65% over k 8192 on one RX 9070 XT PC, level on two and -1.6%
on a fourth, and nothing the app can read tells those PCs apart. So the first
start in a week times both on the PC while the card mines, as RTX 30 times
its launch shapes.

The card mines 30 s uncounted, then four 30 s turns at k 8192, 16384, 16384
and 8192, opening its core again at each change of k (about 1 s each). k 16384
is kept only when it leads by 0.3% on the means, in every turn, and by twice
the gap between the two turns of each k. Anything less keeps k 8192, so a PC
where they are level mines as before. Shorter turns in the order 8192, 16384,
8192, 16384 kept k 16384 on a level PC, and 20 s turns wandered by more than
the bar; these settings gave the same answer every time on five PCs.

The answer is kept per card for a week (k-trials.json). The AMD core takes... (continued)

963 of 963 branches covered (100.0%)

Branch coverage included in aggregate %.

1431 of 1431 relevant lines covered (100.0%)

37.65 hits per line

Jobs
ID Job ID Ran Files Coverage
1 44350093850.1 18 Nov 2026 04:42PM UTC 28
100.0
GitHub Action Run
Source Files on build 44350093850
  • Tree
  • List 28
  • Changed 0
  • Source Changed 0
  • Coverage Changed 0
Coverage ∆ File Lines Relevant Covered Missed Hits/Line Branch Hits Branch Misses
  • Back to Repo
  • Github Actions Build #44350093850
  • be8cbcd2 on github
  • Prev Build on main (#44330987114)
  • Next Build on main (#44408039279)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc