Do you have any insight into the benchmarking that IOG does to set these limits, and how we would evaluate the impact of changing the memory units?
2 Likes
So the way it works roughly is:
- We have a (conceptual) bound on how much time in the block diffusion budget is allocated to script evaluation
- We benchmark all our costing benchmarks to 1 ExCPU = 1 picosecond on a “typical” machine
- The limit can therefore be computed as the number of picoseconds in the script evaluation budget
2 Likes
It’d be really cool if there was some way we could give increased execution units to some percentage of blocks; say, if 1% of blocks were allowed to have this new set of parameters, then we could safely measure the propagation delays on real world topologies by looking at just those blocks, without outsized risk to the protocol overall.
Not sure what the current status of this PCP is, but recent blockchain load seems to be giving some new data that might be helpful on this.
1 Like
Cardano Developer Poll
As part of the ongoing discussion around maxTxExMem and potential parameter adjustments, we’re inviting developers to share their perspective on whether memory units for smart contracts should be increased.
Take the quick survey (less than 2 minutes):
Submit your Feedback
(This survey is being conducted for the Parameter Committee.)
4 Likes