OK, so some NUMA effects, running with numactl --cpunodebind=0 cuts the threaded time by half, thus getting us a speedup but a pretty feeble one.
Bug 1578352 Comment 2 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
OK, so some NUMA effects, running with numactl --cpunodebind=0 cuts the threaded time by half, thus getting us a speedup but a pretty feeble one. But there's still some contention here: reducing the number of threads by using --thread-count=4 brings the time down from about 1750ms to about 1150ms.