[fpedd]

With fewer than six threads the portfolio ran only the incumbent bound and its members one after another, so the first member consumed the whole budget and machines with one to four cores solved none of the eleven challenging Minimalloc instances within 3 s. Rounds now always include the lower-bound rung and run under doubling node budgets starting at max(10k, 2n), since a single dive places all n allocations; try_solve_many reports whether a member stopped early, so an exhausted round proves its best solution optimal (or that none exists) without a second proof round. Nine of the eleven are now proved optimal on one thread in under a second each, with no regression at 8 or 14 threads, and a 128k-allocation decomposable chain proves within the default budget on one thread.

The search re-keys its candidate trees only for placements that pass the bound check and marks placed allocations with a sentinel, raising node throughput 1.1-5.8x with byte-identical search trees under the default heuristics. Decomposition buckets allocations once instead of scanning all of them per band (128k allocations took 15.6 s, now 0.26 s), members are copied once inside the worker, the reorders are built only when greedy misses the lower bound and time remains, and the deadline is checked from the first node. The sweep no longer records zero-length sections at an exit-then-enter instant, so the non-default SECTIONS sort key ('C') counts only non-empty spans. Omni does not use SuperMalloc, so its output is unchanged. The integration test checks proved optima on challenging instances instead of running a 40 s benchmark campaign. Stacked on #75.