There was an error while loading. Please reload this page.
Merge Use into Experiments; Use.md retired Use.md's remaining content (Synopsis skeleton) folded into Experiments as its own section upfront; the rest was already duplicated by the per-path walkthroughs. Sidebar and Home repointed to Experiments.
Experiments: align page style, trim FZG path, add TOC Compression/Decompression sections restyled to bullet-form subcommand notes; markdown tables column-aligned throughout. FZG path trimmed to match LC/Spline economy. Table of contents added upfront.
Add The FZG path: command, why the recenter fix exists, CR/throughput vs stock FZ-GPU Three SDRBench datasets (CESM CLDHGH, Hurricane CLOUDf48, NYX baryon_density), same GPU, measured against an unmodified stock FZ-GPU build. cuSZ's FZG path beats stock's own compression ratio on two of three; throughput trails consistently (their fused kernels vs cuSZ's 3-4 separate launches).
Experiments: de-duplicate the page, hyphenate extents, follow --help on task flags The table-split commit accidentally left a second copy of the Lorenzo and Composing sections; removed. Extents now x-y-z per the CLI synopsis. Task flags follow --help, and note that its advertised --extract is rejected by the parser. lrz-zz dropped for now.
Experiments: split the option table by task, list the -z/-x spellings zip/compress alongside -z and unzip/decompress alongside -x, plus the argv[1] subcommand forms. All twelve spellings verified against the binary.
Experiments: name extents x/y/z, not N/M/K
Experiments: add a command-line composition table; correct the report flags --compare is what prints PSNR, not a report key ('quality' is not one). Time reporting is disabled on both sides, so the examples no longer ask for it.
Add Experiments page: Lorenzo, spline and LC command lines Self-checking round trips for each pipeline. Records two traps found while verifying the commands: --report time fails on the decompress side, and spl-y24 on a 2D field reconstructs wrongly instead of being rejected.