Changelog: note the spot-finding, indexing and GPU memory gains

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VfYvJT5Nb71suJCowRBn5z
2026-08-23 13:16:02 +02:00
parent 312df30463
commit 20edf55063
1 changed files with 2 additions and 0 deletions
@@ -27,6 +27,8 @@ This is an UNSTABLE release. It includes many experimental features, as well as
 * `images_per_file` is chosen from the acquisition when it is not given: a rotation sweep of at most 20000 images goes into a single data file, a grid scan splits on whole fast-axis rows, and stills and serial keep 1000.
 * rugnux: beam-stop detection is substantially faster - the pre-scan reads and accumulates its frames on several threads, and the mask is built in one pass over the image rather than by repeated searches. The shadow it finds is unchanged.
 * rugnux: the `_process.h5` file is written on a thread of its own, so processing no longer waits on it.
+* Faster spot finding and indexing, on the broker and in rugnux: the ring reduction reads four pixels at a time and the FFT indexer gives each direction its own block. The spots found and the lattices indexed are unchanged.
+* GPU memory is no longer allocated for buffers nothing reads, and a worker builds only the engines it uses. A run reserves substantially less device memory.
 * CUDA 12 builds now contain device code for Volta, so the RHEL 8 packages and the portable Linux `.tgz` run on a V100; the CUDA 13 artefacts (RHEL 9, Ubuntu, Windows) remain Turing and newer.
 * Documentation: a security page, and the supported GPU generations and minimum NVIDIA driver version of every released artefact.