Execution Effort Calibration 2. [ExEC2]

[janezpodhostnik]

Why

Execution effort needs to accurately map to actual resources used on transaction execution, otherwise we are not charging fees correctly.

How will we measure success ?

Computation Estimation correctness metric

• Max under-estimation reduced by 2x (from ~64 to at least ~32)
• More customizable TPS loader

DACI

DACI

Role	Assigned
Driver	Technical: @janezpodhostnik , EM: @j1010001
Approver	@Kay-Zee
Consulted	@vishalchangrani
Informed	Flow Protocol engineering

Problem Definition

See: https://flowfoundation.notion.site/Execution-Effort-Calibration-2-Plan-fa80f35d68334eaca2238002a3b9b8fe?pvs=4

Tasks

Preview Give feedback

1. [ExEC2] Use nanosecond resolution for reporting execution time
2. [ExEC2] Change grafana reporting of normalised time per execution effort
3. [ExEC2] Add additional logs to transactions where normalised time per execution effort is far from the mean
4. [ExEC2] Add Recursive least squares method that keeps track of executed transactions
5. [ExEC2] Change synthetic transaction generator, to be able to send transactions to any network
6. [ExEC2] Change synthetic transaction generator, to also allow failed transaction
7. [ExEC2] Add additional transactions to the synthetic transaction generator
8. [ExEC2] Create a database to store transaction execution metrics
9. [ExEC2] Inject a component into the FVM that will upload transaction execution metrics to a database
10. [ExEC2] Collect data from local benchmarks
11. [ExEC2] Collect data from benchnet
12. [ExEC2] Collect testnet data
13. [ExEC2] Data analysis
14. [ExEC2] Write Flip
15. [ExEC2] Collect mainnet data
16. [ExEC2] Finalise data analysis with mainnet data
17. [ExEC2] Update weights
18. [ExEC2] Verify new weights

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.