leonarski_fandClaude Opus 5
2e5c0df0a5
Build Packages / build:rugnux:aarch64 (cross) (push) Successful in 8m58s
Build Packages / build:windows:nocuda (push) Successful in 17m17s
Build Packages / build:rugnux-tgz (x86_64) (push) Successful in 18m48s
Build Packages / build:windows:cuda (push) Successful in 19m38s
Build Packages / build:viewer-tgz:cpu (push) Successful in 20m51s
Build Packages / build:rpm (rocky9_nocuda) (push) Successful in 22m30s
Build Packages / build:viewer-tgz:cuda (push) Successful in 23m33s
Build Packages / build:rpm (rocky8_nocuda) (push) Successful in 27m3s
Build Packages / build:rugnux:windows (push) Successful in 11m35s
Build Packages / build:rpm (ubuntu2204_nocuda) (push) Successful in 28m49s
Build Packages / build:rpm (ubuntu2404_nocuda) (push) Successful in 20m3s
Build Packages / build:rpm (rocky9_sls9) (push) Successful in 22m30s
Build Packages / build:rpm (rocky8_sls9) (push) Successful in 26m57s
Build Packages / build:rpm (rocky9) (push) Successful in 25m1s
Build Packages / build:rpm (rocky8) (push) Successful in 29m28s
Build Packages / build:rpm (ubuntu2404) (push) Successful in 23m36s
Build Packages / Generate python client (push) Successful in 42s
Build Packages / Create release (push) Skipped
Build Packages / Build documentation (push) Successful in 1m54s
Build Packages / XDS test (durin plugin) (push) Successful in 11m11s
Build Packages / build:rpm (ubuntu2204) (push) Successful in 28m9s
Build Packages / DIALS test (push) Successful in 26m17s
Build Packages / XDS test (JFJoch plugin) (push) Successful in 10m33s
Build Packages / XDS test (neggia plugin) (push) Successful in 8m45s
Build Packages / Unit tests (push) Successful in 1h50m12s
The three step-4b passes - the global decay slope, the per-batch relative-B and the radiation-damage monitor - were plain serial walks over the eighty-byte observation records, a dozen of them, each reading twenty bytes of each record and each allocating and zeroing a pair of per-group vectors. The decay fit alone runs seven of those passes (three slope fits, four held-out scores) on one thread while the workers and the GPU idle. Give them the treatment the correction surfaces below them already have: pull the fields they read out of the records once into a compact array in fulls order, and keep a second copy of the same terms in ASU-group order (stable counting sort) so a thread can own whole groups. The per-group inverse-variance references then run on all threads and still add each group's terms in fulls order, and the remaining scalar reductions keep their serial order over the compact array - so every sum is formed from the same terms in the same order as before, and the numbers are unchanged rather than merely close. The build is shared by the three curve fits and the three held-out scores of the relative-B pass instead of being repeated per call. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>