Sequenced state transitions😎 Discussed often recently, where validator sets responsible for state with many transitions can optimize execution. Thought I'd try an implementation & test my theory. 16 groups, 128 nodes + usual node spec = ~31k swaps/s 🥳 $xrd #crypto #radix

Some clarity: Substate X is pool state Lots of transactions want to swap on the pool Validator set A is responsible for substate X Validator set A determines locally the order that the related transactions will mutate substate X State changes to X can be accumulated rather than being applied individually. This greatly reduces I/O and memory use, which allows more time actually executing. Its tricky though because you have to take into consideration various issues such as transactions that fail, timeout or become latent due to some external validator group issue. Handling those cases is the complex piece to ensure that the state retains integrity at the end of the sequence.

#Radix 1.8m TPS wtf!! Just need a partnership with Visa or MasterCard now.

These is 1.860.000 tpm impressive.

yeah about that its still a bit sketchy, needs a lot more work but good enough for now as a demo

Y'all out there, wear your seatbelts because we're going fast 🚄

Dan is this fast?

🤯🤯

Impressive achievement again! Usual questions: Finality still ~3.5s in this scenario? And *real* TPS 62k ish?

You can see it on the right hand window which is one of the validators I'm poking commands into. Its ~2.8s so a little faster because of the sequencer. There's also other work I've done since the 9k one which helps finality a little too. Whats interesting is that sequenced throughput actually approaches basic transfer throughput the more you load it. So the *real* TPS as you put it is also about 30-35k.

Haven't a Danny what your on about, but seems epic.