Distance pallet benchmark

assigned to @bgallois

J'avais commencé le benchmark de distance mais n'ai plus avancé depuis un moment, sur la branche tuxmain/benchmark-distance que je viens de pousser. Comme il y a déjà le mock et certaines fonctions tu peux reprendre le code si c'est utile.

added 1 commit

• c48cb991 - wip

Compare with previous version

added 1 commit

• 1a5827be - wip

Compare with previous version

added 1 commit

• 08fac5f8 - wip

Compare with previous version

added 1 commit

• c7ec9c4f - add benchmarks for runtime

Compare with previous version

added 1 commit

• eb80b819 - add benchmarks for runtime

Compare with previous version

added 1 commit

• 6c9538f0 - add benchmarks for test

Compare with previous version

added 1 commit

• ca2de481 - use our weights

Compare with previous version

added 5 commits

• 1157b8ad - 1 commit from branch nodes/rust:master
• eecb80cf - add pallet distance benchmarks
• d3900255 - add benchmarks for runtime
• b759bf98 - add benchmarks for test
• bbf95632 - use our weights

Compare with previous version

added 1 commit

• 859e8204 - use our weights

Compare with previous version

added RN-runtime label

changed milestone to %runtime-700

added 1 commit

• 5ecf2fe7 - use our weights

Compare with previous version

added 1 commit

• 2578fad1 - fix runtime-benchmarks feature for testing

Compare with previous version

added 1 commit

• f7221366 - fix runtime-benchmarks feature for testing

Compare with previous version

added 1 commit

• 7c308146 - fix runtime-benchmarks feature for testing

Compare with previous version

added 1 commit

• baef83b3 - add digest decoding complexity

Compare with previous version

added 1 commit

• b2ab1db4 - add digest decoding complexity

Compare with previous version

• Added benchmarks for the distance pallet (for both testing and runtime).
• Included our weights in the runtime.
• Calculated provisional weights using my personal machine (to be recomputed on the target machine).

Note:
One technical point is the author check at https://git.duniter.org/nodes/rust/duniter-v2s/-/blob/master/pallets/distance/src/lib.rs#L228 that is implemented at https://github.com/paritytech/substrate/blob/033d4e86cc7eff0066cd376b9375f815761d653c/frame/authorship/src/lib.rs#L80-L93 and then at https://github.com/paritytech/substrate/blob/033d4e86cc7eff0066cd376b9375f815761d653c/frame/babe/src/lib.rs#L493-L507. Therefore, the complexity of this check depends on the number of items and the ordering of the items.

Based on my research and testing, in our configuration, it seems that only BABE and GRANDPA use the Digest system. The number of items seems to be always one, or the BABE digest is the first when the extrinsic call is performed. Therefore, I have kept the number at one during the benchmark and have not added another complexity parameter.

requested review from @HugoTrentesaux

marked this merge request as ready

resolved all threads

added 1 commit

• 45b62973 - manual formatting

Compare with previous version

Did not really understand this Digest system. I mark this MR as "approved" but maybe @tuxmain would like to review more carefully.

requested review from @tuxmain and removed review request for @HugoTrentesaux

approved this merge request

resolved all threads

added 17 commits

• 45b62973...b793b455 - 11 commits from branch nodes/rust:master
• 1d5cd029 - add pallet distance benchmarks
• 0adcd132 - add benchmarks for runtime
• fc138f7c - add benchmarks for test
• 606e617a - use our weights
• dbb186db - fix runtime-benchmarks feature for testing
• 6cec1103 - add digest decoding complexity

Compare with previous version

added 1 commit

• 41bdac4a - fix after merge

Compare with previous version

merged

mentioned in commit 89c4fdb9

mentioned in issue #126 (closed)