Does anyone want to challenge eMunie's claim in this bake-off? BitShares' Stan/Dan? Come-from-Beyond?
I need numbers for
payments-only performance of eMunie then.
I'm not sure what you mean by payments-only, isn't that all this test is going to do? We cant
turn off all the additional transactional features in order to achieve higher throughput, nor would I want to. Other projects may wish to do that, but I want to have a near as possible production eMunie configuration so that its a truer test of what real world performance can be.
Anyway, for this "bake-off" eMunie will most likely run in a single partition configuration as there is still some final work and testing to do in order to operate a multi-partition network. If the bake-off date is indeed moved though, we should be at a point where a multi-partition network could be configured.
With the machine specs outlined in the OP, and a single partition config, eMunie should be able to easily sustain 200-300 tx/s across 20 nodes for 100 minutes. Taking the average of 250 tx/s, our test result should be somewhere around 1.5M transactions after 100 minutes. If the OP wants decides to do peak testing too, 3-5k tx/s should be possible over a short duration, but this bake-off really needs sustained load capability, as the final transaction count is what matter.
If we are at a stage where we can operate a multi-partition network, then that throughput
greatly increases up to an order of 10x for 20 partitions!