update weights #6025

coderobe · 2022-09-19T21:41:16Z

new weights as requested by @shawntabrizi

coderobe · 2022-09-19T21:51:16Z

this was run off d17f062

shawntabrizi · 2022-09-20T19:57:33Z

runtime/kusama/src/weights/frame_benchmarking_baseline.rs

+		Weight::from_ref_time(105_000 as u64)
 	}
 	/// The range of component `i` is `[0, 1000000]`.
 	fn subtraction(_i: u32, ) -> Weight {
-		Weight::from_ref_time(125_000 as u64)
+		Weight::from_ref_time(112_000 as u64)


@koute @ggwpez strange to me that these end in 000?

What are the chances if the precision is down to a single unit

I think the increased precision only applied to the linear component, or @koute ?

that makes sense

Correct, the increased precision is only for the component weights, as that's where it can make a difference.

shawntabrizi · 2022-09-20T19:58:28Z

runtime/kusama/src/weights/frame_election_provider_support.rs

+		Weight::from_ref_time(5_800_910_000 as u64)
+			// Standard Error: 141_466
+			.saturating_add(Weight::from_ref_time(5_793_269 as u64).saturating_mul(v as u64))
+			// Standard Error: 14_463_035
+			.saturating_add(Weight::from_ref_time(1_592_527_344 as u64).saturating_mul(d as u64))
 	}
 	/// The range of component `v` is `[1000, 2000]`.
 	/// The range of component `t` is `[500, 1000]`.
 	/// The range of component `d` is `[5, 16]`.
 	fn phragmms(v: u32, _t: u32, d: u32, ) -> Weight {
-		Weight::from_ref_time(0 as u64)
-			// Standard Error: 79_000
-			.saturating_add(Weight::from_ref_time(14_480_000 as u64).saturating_mul(v as u64))
-			// Standard Error: 6_844_000
-			.saturating_add(Weight::from_ref_time(2_525_332_000 as u64).saturating_mul(d as u64))
+		Weight::from_ref_time(4_696_655_000 as u64)
+			// Standard Error: 150_968
+			.saturating_add(Weight::from_ref_time(5_769_927 as u64).saturating_mul(v as u64))
+			// Standard Error: 15_434_495
+			.saturating_add(Weight::from_ref_time(1_820_652_967 as u64).saturating_mul(d as u64))


big changes here

The worst-case improved for both runtimes, rococo is unchanged.:

It does look like this weight was affected by the zero base weight issue that my PR fixed; looks correct to me. As @ggwpez pointed out this actually lowered the weight final weight since the negative base weight isn't rounded up to zero anymore (so we were overestimating how costly this extrinsic was when the components are not zero)

shawntabrizi · 2022-09-20T19:59:45Z

/cmd queue -v RUST_LOG=debug -c bench-bot $ runtime kusama-dev pallet_balances

command-bot · 2022-09-20T19:59:49Z

@shawntabrizi https://gitlab.parity.io/parity/mirrors/polkadot/-/jobs/1864571 was started for your command "$PIPELINE_SCRIPTS_DIR/bench-bot.sh" runtime kusama-dev pallet_balances. Check out https://gitlab.parity.io/parity/mirrors/polkadot/-/pipelines?page=1&scope=all&username=group_605_bot to know what else is being executed currently.

Comment /cmd cancel 41-68d77c72-0436-41d4-b2c1-c58b40841f82 to cancel this command or /cmd cancel to cancel all commands in this pull request.

command-bot · 2022-09-20T20:05:39Z

@shawntabrizi Command "$PIPELINE_SCRIPTS_DIR/bench-bot.sh" runtime kusama-dev pallet_balances has finished. Result: https://gitlab.parity.io/parity/mirrors/polkadot/-/jobs/1864571 has finished. If any artifacts were generated, you can download them from https://gitlab.parity.io/parity/mirrors/polkadot/-/jobs/1864571/artifacts/download.

ggwpez · 2022-09-20T21:08:17Z

Interesting. There are some big increases, but also a lot of decreases.
Remark jumped to 1.46ms, but it seems consistent across the runtimes. Westend was not updated @coderobe?

SWC-compare link link and PDF for posterity.

coderobe · 2022-09-20T21:09:45Z

westend wasn't ready when i opened the PR - it is now, i'll add the files.

shawntabrizi · 2022-09-20T21:11:04Z

runtime/kusama/src/weights/pallet_collective_technical_committee.rs

+			.saturating_add(T::DbWeight::get().reads(3 as u64))
 			.saturating_add(T::DbWeight::get().reads((1 as u64).saturating_mul(p as u64)))
-			.saturating_add(T::DbWeight::get().writes(2 as u64))
+			.saturating_add(T::DbWeight::get().writes(3 as u64))


So the weights were outdated somehow

shawntabrizi · 2022-09-20T21:22:23Z

There are some changes here we should look into, but i feel pretty confident after looking over this, that the changes that @koute made are solid, and probably we need to hunt separately what is going on

Thanks @coderobe

coderobe · 2022-09-20T21:23:26Z

doesn't compile, however

shawntabrizi · 2022-09-20T21:41:39Z

Looks like an error in the template, should be easy to resolve, will look into it

ggwpez · 2022-09-20T21:50:14Z

Looks like an error in the template, should be easy to resolve, will look into it

Could also just be missing deps in runtime/kusama/constants/Cargo.toml.
PS I think we can and should get rid of that sub-crate and just put it into the runtime.

ggwpez · 2022-09-27T13:52:37Z

Regarding the merge conflicts: its since we renamed anonymous to pure. I will do a git merge -X theirs origin/master.

Signed-off-by: Oliver Tale-Yazdi <[email protected]>

ggwpez

The large changes are most likely from the recent weight calculation changes. We are investigating them in Substrate.
Should be more precise now 😄

ggwpez · 2022-09-27T15:04:10Z

runtime/kusama/constants/src/weights/block_weights.rs

@@ -1,28 +1,27 @@
-// This file is part of Substrate.
+// Copyright 2017-2022 Parity Technologies (UK) Ltd.


I added the correct file header manually. All future MRs should have them automatically #5984

coderobe added 2 commits September 19, 2022 23:38

rococo: update weights

1f886b9

polkadot: update weights

2aeb4d9

coderobe requested review from shawntabrizi and ggwpez September 19, 2022 21:41

github-actions bot added the A3-in_progress Pull request is in progress. No review needed at this stage. label Sep 19, 2022

coderobe added B0-silent Changes should not be mentioned in any release notes C1-low PR touches the given topic and has a low impact on builders. D3-trivial 🧸 PR contains trivial changes in a runtime directory that do not require an audit. labels Sep 19, 2022

kusama: update weights

d6bf724

shawntabrizi reviewed Sep 20, 2022

View reviewed changes

westend: update weights

7c316c1

shawntabrizi reviewed Sep 20, 2022

View reviewed changes

fmt

989488c

coderobe marked this pull request as ready for review September 20, 2022 21:12

coderobe added A0-please_review Pull request needs code review. and removed A3-in_progress Pull request is in progress. No review needed at this stage. labels Sep 20, 2022

shawntabrizi added 2 commits September 20, 2022 18:04

add deps

610fb13

fix warnings

ac5857f

shawntabrizi mentioned this pull request Sep 21, 2022

New Weights for All Pallets paritytech/substrate#12325

Merged

ggwpez added 2 commits September 27, 2022 15:53

Merge remote-tracking branch 'origin/master' into coderobe/weights-st

7c6fc53

Add file_header.txt to block weight files

89e254f

Signed-off-by: Oliver Tale-Yazdi <[email protected]>

ggwpez approved these changes Sep 27, 2022

View reviewed changes

coderobe mentioned this pull request Sep 30, 2022

update weights #6087

Closed

coderobe closed this Jan 16, 2023

coderobe deleted the coderobe/weights-st branch January 16, 2023 14:42

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

update weights #6025

update weights #6025

coderobe commented Sep 19, 2022 •

edited

Loading

coderobe commented Sep 19, 2022

shawntabrizi Sep 20, 2022 •

edited

Loading

ggwpez Sep 20, 2022

shawntabrizi Sep 20, 2022

koute Sep 21, 2022

shawntabrizi Sep 20, 2022

ggwpez Sep 20, 2022

koute Sep 21, 2022

shawntabrizi commented Sep 20, 2022

command-bot bot commented Sep 20, 2022 •

edited

Loading

command-bot bot commented Sep 20, 2022

ggwpez commented Sep 20, 2022

coderobe commented Sep 20, 2022

shawntabrizi Sep 20, 2022

ggwpez Sep 20, 2022

shawntabrizi commented Sep 20, 2022 •

edited

Loading

coderobe commented Sep 20, 2022

shawntabrizi commented Sep 20, 2022

ggwpez commented Sep 20, 2022 •

edited

Loading

ggwpez commented Sep 27, 2022

ggwpez left a comment

ggwpez Sep 27, 2022

		@@ -1,28 +1,27 @@
		// This file is part of Substrate.
		// Copyright 2017-2022 Parity Technologies (UK) Ltd.

update weights #6025

update weights #6025

Conversation

coderobe commented Sep 19, 2022 • edited Loading

coderobe commented Sep 19, 2022

shawntabrizi Sep 20, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

shawntabrizi commented Sep 20, 2022

command-bot bot commented Sep 20, 2022 • edited Loading

command-bot bot commented Sep 20, 2022

ggwpez commented Sep 20, 2022

coderobe commented Sep 20, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

shawntabrizi commented Sep 20, 2022 • edited Loading

coderobe commented Sep 20, 2022

shawntabrizi commented Sep 20, 2022

ggwpez commented Sep 20, 2022 • edited Loading

ggwpez commented Sep 27, 2022

ggwpez left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

coderobe commented Sep 19, 2022 •

edited

Loading

shawntabrizi Sep 20, 2022 •

edited

Loading

command-bot bot commented Sep 20, 2022 •

edited

Loading

shawntabrizi commented Sep 20, 2022 •

edited

Loading

ggwpez commented Sep 20, 2022 •

edited

Loading