The endpoint sweep of Section 4 as a word RAM program: read the word, sort the jobs by start time ($O(n log n)$), build the arrays of the sorted instance and the order of the due dates, then sweep the $2n$ events — the start of a job takes a free slot (or a fresh one, doubling the mask range $MK = 2^next$), the due date frees it — updating a table with one entry for every set of occupied slots and every weight up to $W$, by a flat pass of $MK · (W+1)$ cells per event. The number of slots is at most the width $ω$, so every event costs $O(2^ω · (W+1))$ and the run is within $c · (W+1) · 2^ω · (n+1) + c · sortCost$. The domain restricts to positive processing times, the paper's standing assumption for the sweep.