Cards buy time, not throughput.

A shooting day is a queue. Storage fills at duty × data rate and drains at offload ÷ passes. If the fill exceeds the drain the day is not sustainable, and no number of cards makes it so — capacity is a buffer, not a pipe.

The engine is free and needs no account. Two bodies, seven cards, a perfectly ordinary reader. The pool is 954 GiB, it fills at 144 MB/s and drains at 100 MB/s once two copies are counted, so storage accumulates at 43.8 MB/s and the whole pool is gone 6.5 h into a ten-hour day. Doubling the cards moves that to 13 h and changes the deficit not at all; what fixes it is 288 MB/s of pipe, or rolling 34.8% of the time instead of half. And there are two questions here, not one — the same day that passes on rate can still need 5.6 hours of copying in the two hours after wrap. Every lane ships a worked example that costs nothing. Writing the report needs credits.