Benchmarking xcm instructions Weight #1012
Conversation
gitofdeepanshu
added
runtime
gitofdeepanshu
added
shibuya
|
/bench shibuya-dev astar_xcm_benchmarks::generic |
|
Benchmarks job is scheduled at https://github.com/AstarNetwork/Astar/actions/runs/6536053493. |
|
Benchmark job failed. |
|
/bench shibuya-dev astar_xcm_benchmarks |
|
Benchmarks job is scheduled at https://github.com/AstarNetwork/Astar/actions/runs/6542412489. |
|
Benchmarks have been finished. |
gitofdeepanshu
changed the title
|
/bench shibuya-dev astar_xcm_benchmarks::generic,astar_xcm_benchmarks::fungible::assets,astar_xcm_benchmarks::fungible::balances |
|
Benchmarks have been finished. |
|
/bench shibuya-dev astar_xcm_benchmarks::generic |
|
Benchmarks job is scheduled at https://github.com/AstarNetwork/Astar/actions/runs/6574576146. |
|
Benchmarks have been finished. |
|
/bench shibuya-dev astar_xcm_benchmarks::generic,astar_xcm_benchmarks::fungible::assets,astar_xcm_benchmarks::fungible::balances |
|
Benchmarks job is scheduled at https://github.com/AstarNetwork/Astar/actions/runs/6585607290. |
|
Benchmarks have been finished. |
|
/bench shibuya-dev astar_xcm_benchmarks::generic,astar_xcm_benchmarks::fungible::assets,astar_xcm_benchmarks::fungible::balances |
|
Benchmarks job is scheduled at https://github.com/AstarNetwork/Astar/actions/runs/6589059865. |
|
Benchmarks have been finished. |
Minimum allowed line rate is |
| benchmarks_instance_pallet! { | ||
| where_clause { where | ||
| < | ||
| < | ||
| T::TransactAsset | ||
| as | ||
| Inspect<T::AccountId> | ||
| >::Balance | ||
| as | ||
| TryInto<u128> | ||
| >::Error: sp_std::fmt::Debug, | ||
|
|
||
| < T::TransactAsset | ||
| as | ||
| Inspect<T::AccountId> | ||
| >::Balance : Into<u128> | ||
|
|
||
| } |
There was a problem hiding this comment.
Why use old v1 benchmarks if new v2 has been available for many months?
There was a problem hiding this comment.
changed to v2.
| pub mod assets; | ||
| pub mod balances; |
There was a problem hiding this comment.
Are two completely separate modules really needed?
Seems like lots of copy/paste, for mock.rs and mod.rs
There was a problem hiding this comment.
I wanted our pallet to have consistent style with the original pallet so I used the same format. We can also merge them into one but that would mean changing the variable names and name of benchmarks and some logic which will be different from the original pallet.
Let me know what you think.
|
/bench shibuya-dev astar_xcm_benchmarks::generic,astar_xcm_benchmarks::fungible::assets,astar_xcm_benchmarks::fungible::balances |
|
Benchmarks job is scheduled at https://github.com/AstarNetwork/Astar/actions/runs/6728742049. |
|
Benchmarks have been finished. |
Pull Request Summary
Currently all xcm instructions occupy same weight, this PR changes that by changing
FixedWeightboundstoWeightInfoBoundswhich benchmarks all xcm instructions for accurate weight.Benchmarking pallets are inspired from
pallet_xcm_benchmarkand adjustments related to Shibuya has been made to benchmarks.Architecture
Just like the original pallet, instructions are divided between
genericandfungibleinstructions, and each pallet's config is configurable to define how the benchmarks will be run.Furthermore, for
fungibleinstrctions, separate pallets for assets of typebalancesandassetswere added to keep the naming of benchmarks and config consistent and straightforward.Weight Calculations
For generic instructions we have taken the worst case scenarios that could occur.
For fungible instructions
We benchmark every instruction for 1 asset of
pallet_assetsand 1 asset ofpallet_balancesand take the maximum weight consumed between them. For multiple assets, we multiply that weight with the number of assets.Weight = max_weight_consumption( one pallet_balances token , one pallet_assets token) for that instruction * number of assets