Ackren · Two folds · PSRAM
Quarto, two folds
Roughly 200 MB against a card. The middle floor, and the one most embedded targets can actually meet.
The floor
PSRAM and a card
Quarto expects external PSRAM for working state and an SD card for the artefact. That is a board most embedded teams already have, or can reach without a redesign, which is why this is the format we expect the first regulated deployment to use.
Roughly 200 MB buys fingerprints wide enough to put the silent false accept rate at one in 16.7 million. At realistic device lifetimes that is a number a certification body can be shown without an argument attached.
Choosing between Quarto and Octavo
Query volume, not ambition
The decision is arithmetic rather than taste. Estimate lookups per device lifetime, multiply by the false accept rate for each format, and see which answer you are willing to write down in a hazard analysis. If that number is uncomfortable at one in 4,096, you need Quarto.
Nothing else changes between them. The grammar is the same, the packs are the same, the derivation log is the same. Only the width of the fingerprints moves.
The card is a real dependency
Quarto needs storage that Octavo does not. If your enclosure, cost model or certification path rules out a card, that decision is made for you and it is made before the software conversation starts.