#dusk $DUSK @Dusk been staring at Dusk's engineering notes instead of the deck for once, and this rabbit hole ended up somewhere I didn't expect.
block size is capped at 1Mb. dusk says that's roughly 250 Phoenix (shielded) transactions per block, done on purpose to stop certain attacks.
then I remembered fee data I pulled a while back — the explorer's own stats page puts it at 8,639 blocks with a 10.0 sec avg block time. clean number, no need to do the division myself this time.
250 tx a block, one block every ~10 sec.. that's a ceiling of ~25 shielded tx/sec. over a full day, best case, ~2.16 million Phoenix transactions.
then I checked DTC — the Depository Trust Company, part of the plumbing under US securities settlement. processes a little over 1.3 million transactions a day.
wait.
Dusk's theoretical ceiling is higher than that. on paper.
hold up though.. that comparison is sneaky. DTC's 1.3M is real, sustained, happens every day. Dusk's 2.16M assumes every block hits exactly 250 Phoenix tx, zero Moonlight competing, zero underfilled blocks. that's never been observed.
and the 250-tx number itself got shakier the more I sat with it. it's data size — 1MB ÷ avg tx size. but Phoenix carries zk-proof verification, Moonlight doesn't. if they're competing for the same block, does 1MB give out first, or does something else — proof load, execution time — hit the wall sooner?
went digging on that last part instead of just leaving it. turns out Dusk added a second cap — a configurable block gas limit, explicitly so the network "adapts under load without bottlenecks." their words. someone already suspected gas, not size, might be the real ceiling.
and they've run named stress environments for this since 2024 — Lunare, Nocturne, ITN.
so it's not that nobody's tested it. the tooling's there. nobody's published the number — no Phoenix TPS under load, no proof verification time, no answer on which cap bites first.
is an untested ceiling still something you build institutional trust on, or just math until someone publishes the receipts?
#zkProofs
block size is capped at 1Mb. dusk says that's roughly 250 Phoenix (shielded) transactions per block, done on purpose to stop certain attacks.
then I remembered fee data I pulled a while back — the explorer's own stats page puts it at 8,639 blocks with a 10.0 sec avg block time. clean number, no need to do the division myself this time.
250 tx a block, one block every ~10 sec.. that's a ceiling of ~25 shielded tx/sec. over a full day, best case, ~2.16 million Phoenix transactions.
then I checked DTC — the Depository Trust Company, part of the plumbing under US securities settlement. processes a little over 1.3 million transactions a day.
wait.
Dusk's theoretical ceiling is higher than that. on paper.
hold up though.. that comparison is sneaky. DTC's 1.3M is real, sustained, happens every day. Dusk's 2.16M assumes every block hits exactly 250 Phoenix tx, zero Moonlight competing, zero underfilled blocks. that's never been observed.
and the 250-tx number itself got shakier the more I sat with it. it's data size — 1MB ÷ avg tx size. but Phoenix carries zk-proof verification, Moonlight doesn't. if they're competing for the same block, does 1MB give out first, or does something else — proof load, execution time — hit the wall sooner?
went digging on that last part instead of just leaving it. turns out Dusk added a second cap — a configurable block gas limit, explicitly so the network "adapts under load without bottlenecks." their words. someone already suspected gas, not size, might be the real ceiling.
and they've run named stress environments for this since 2024 — Lunare, Nocturne, ITN.
so it's not that nobody's tested it. the tooling's there. nobody's published the number — no Phoenix TPS under load, no proof verification time, no answer on which cap bites first.
is an untested ceiling still something you build institutional trust on, or just math until someone publishes the receipts?
#zkProofs