[perf+docs]: p-value path ~70-78x faster + optimization backlog (optimization-backlog) - #38

Open

haoyu-zc wants to merge 5 commits into

refacotring-07-16-2026from

pr-f-optimization-backlog

Commits on Jul 17, 2026

[docs]: Add ranked GPU optimization backlog + captured baseline (optimization-backlog)

Haoyu Zhang
authored and
haoyu-zc
committed
9fa8cfc
[docs]: Mark deferred profiling work honestly in the backlog (codex)

Haoyu Zhang
authored and
haoyu-zc
committed
ad5d261
[perf]: Parallelize p-value permutations across comparisons (H2)
Haoyu Zhang
committed
3ba5df1
[perf]: Coalesced per-thread local contingency in the p-value kernel (H2b)
Haoyu Zhang
committed
5ba8362
[docs]: Add GPU optimization results report (H2 + H2b, ~70-78x p-value speedup)
Haoyu Zhang
committed
fced7db