Questions tagged [benchmarking]
The benchmarking tag has no usage guidance.
113
questions
0
votes
1
answer
19
views
Do struct allocations' weight consumption depend linearly on input data length?
I am writing some benchmarking code and I came across a situation where I had to benchmark a few let bindings.
let sig = ...;
let data : Vec<u8> = vec![...];
// data is actually variable. Ranges ...
- 3,612
0
votes
0
answers
17
views
How does the benchmarking code add_benchmarks! work on the runtime?
I am trying to understand the runtime-benchmarks gated code on the runtime/src/lib.rs and I realize that we need to "list" and "add" benchmarks that we want to perform : https://...
- 502
1
vote
1
answer
18
views
How to generate database weights?
As a chain grows the DB operations should get costlier. Is there a standard procedure that outlines how to generate these weights? Or should we continue using the default substrate weights for this ...
- 3,612
2
votes
1
answer
24
views
Benchmarking an extrinsic that validates a signature
I have an extrinsic in one of my pallets that looks validates a signature over a message, which is separate from the actual extrinsic being submitted.
The code can be simplified to look like the below
...
- 418
2
votes
1
answer
30
views
Benchmarking in a cusom pallet requires a storage put operation in Staking pallet
Let's say I have a pallet_x. In this pallet i have an extrinsic where I get the active_era using the below code :
Staking::<T>::active_era()
.map(|e| e....
- 674
3
votes
1
answer
68
views
Quadratic complexity in Benchmarking
I am trying to understand if quadratic complexity is bad (gives inaccurate results) when used for benchmarking:
I have a pallet whose weight would ideally depend on x * y where x and y are lengths of ...
- 3,612
1
vote
1
answer
20
views
How to account for ensure_signed calls?
Let's say our pallet dispatchable returns DispatchResultWithPostInfo, where we are trying to manually calculate the weight, albeit with helpers obtained from benchmarking, but that's not relevant to ...
- 3,612
4
votes
1
answer
29
views
Can we use std crates in benchmarking?
For benchmarking I can see a need to use random number generators for maybe verification of a random storage or such. Do we have utilities provided by the frame_benchmarking framework itself or is it ...
- 3,612
0
votes
1
answer
25
views
Who are whitelisted_callers?
On benchmarks one can often see whitelisted_caller() being used to test extrinsics and I wonder, how do they bypass the need for fee, and even pay for extrinsics like transfer_keep_alive for ...
- 3,612
2
votes
1
answer
40
views
How to benchmark frame-system on parachain?
After upgrading the Substrate to polkadot-v0.9.43.
I got the following error while running the runtime benchmarks.
2023-07-21 19:55:56 Starting benchmark: frame_system::set_code
Error: Input("...
- 8,678
3
votes
1
answer
30
views
Error: "Did not find the benchmarking metadata" after building runtime for benchmarking
After upgrading our chain to polkadot 0.9.38 I managed to build the benchmarking using cargo build --release --features runtime-benchmarks but when I try to benchmark any pallet after the build I ...
- 273
7
votes
1
answer
140
views
How does the benchmark calculate the proof size?
I have no idea why this call uses such a lot proof size (around 4MB).
It looks like it doesn't change too many storage items.
https://github.com/darwinia-network/darwinia-messages-substrate/blob/...
- 8,678
5
votes
1
answer
39
views
frame-benchmarking::v2 doesn't work for force_origin
In Config of my pallet I have:
type FilteredOrigin: EnsureOriginWithArg<
Self::RuntimeOrigin,
SomeEnum,
Success = Self::AccountId
>;
And there is an ...
- 217
5
votes
0
answers
32
views
Benchmarks not getting skipped with Err(BenchmarkError::Skip)
I am using the pallet_xcm_benchmarks, in the config for pallet_xcm_benchmarks::generic,
I have used
impl pallet_xcm_benchmarks::generic::Config for Runtime {
fn worst_case_asset_exchange() -> ...
- 51
4
votes
1
answer
74
views
Missing XCM benchmark `reserve_asset_deposited`
I'm upgrading our parachain based on Polkadot 0.9.40 here. Now in the process of benchmarking xcm operations, I notice there is no benchmark for reserve_asset_deposited in XCM fungible benchmarks! I ...
- 493
4
votes
1
answer
32
views
Associated types in benchmark `where` bound
I'm trying to constrain the value of an associated type of a trait inside a benchmark's where bound, like this:
#[benchmarks(where T: currency::Config<Balance = u128>)]
pub mod benchmarks {
...
- 362
3
votes
1
answer
47
views
Can't write to StorageMap when benchmarking, EncodeLike is not implemented for any type even if it's the same in pallet
I have the following StorageMap defined in my pallet:
#[pallet::storage]
pub(super) type Signatures<T: Config> = StorageMap<
_,
Blake2_128Concat,
T::Hash,
BoundedBTreeMap<...
- 1,053
3
votes
2
answers
26
views
Does substrate offer benchmarking utilities in outer node?
I have some heavy computation done off-chain (in outer node) and i want to benchmark those functions, does substrate offer tools for doing it or should i rely on a third party library to do that?
- 1,053
3
votes
1
answer
47
views
How to create a bunch of key pairs when benchmarking?
I want to benchmark verifying sr25519 signatures and for that I need to create a bunch of key pairs, sign some data, then benchmark the verification process, however using sp_core::sr25519::Pair is ...
- 1,053
3
votes
1
answer
47
views
Benchmark failing extrinsics
Is there a way to benchmark an extrinsic whose expected behavior is to trigger an error?
I stumbled on a project where we have two instances of the same pallet. Instance1 will proceed normally for a ...
3
votes
3
answers
64
views
Custom benchmarks for external pallets
We use some external pallets that are highly configurable. The weight of the extrinsics in these pallets is highly dependent on the callback supplied in the config, and the existing benchmarks don't ...
- 362
6
votes
1
answer
105
views
cannot create module: instance allocation for this module requires 74376 bytes which exceeds the configured maximum of 65536 bytes
Try in to benchmark node pallets after Polkadot upgrade from 0.9.36 to 0.9.38 and get cannot create module: instance allocation for this module requires 74376 bytes which exceeds the configured ...
- 188
2
votes
1
answer
54
views
Estimated proof size less than measured
I was running some benchmark, and noticed that the estimated proof size was less than the measured size, despite MaxEncodedLen being used in this crate. The estimation of 0 is clearly wrong, but this ...
- 362
2
votes
1
answer
58
views
What is Dev Mode? And how can I use it?
I heard there is something called Dev Mode in Substrate for pallet development.
What is Dev Mode? And how can I use it?
- 7,705
6
votes
1
answer
63
views
Benchmarking multiple instances
We have two instances of a pallet with different runtime configurations. How can we run the benchmarks separately to generate two distinct sets of weights?
As far as I can tell this is not possible ...
- 320
6
votes
1
answer
62
views
Proof Skipped in benchmark
We're updating our weights to weights 2.0 but we're getting a lot of Proof Skipped lines in the output. What is the impact of this, and how do we fix it?
Example output:
/// Proof Skipped: Escrow ...
- 362
3
votes
1
answer
52
views
`Currency::create_deposit` of `pallet_balance` fail on actual runtime but passes on mock rutime
The benchmark is written as:
one_extrinsic {
let caller: <T as frame_system::Config>::AccountId = make_account::<T>(1);
assert_eq!(
<T as crate::Config&...
- 646
0
votes
1
answer
145
views
1010: Invalid Transaction: Transaction would exhaust the block limits
How to specify limit weight for extrinsic?
- 511
0
votes
0
answers
41
views
Error: Input Runtime benchmarking wasn't enabled when building the node
I'm following along this benchmark docs to run benchmarking on my Substrate node (dependencies on branch = "polkadot-v0.9.25") with the command cargo build --release --features runtime-...
- 71
0
votes
0
answers
21
views
Use sightglass for wasmtime benchmarking
Did anyone ever successfully use sightglass for benchmarking of Substrate in wasmtime?
I'm trying to build the following sightglass benchmarking suite / Dockerfile:
RUN echo 'APT::Install-Suggests &...
- 565
0
votes
1
answer
37
views
How to benchmark huge bounded vectors?
I want to have a queue system in a pallet using a storage value and a bounded vector.
/// Data related to tranmission queues.
#[pallet::storage]
#[pallet::getter(fn at_block_queue)]
pub type ...
- 177
0
votes
1
answer
25
views
How many (order of magnitude) Parachains can run on a Relaychain?
Does anyone have data/experience that indicates how many Substrate Parachains can run under one Substrate Relaychain?
I understand this will be hardware specific, so it will help if you can give some ...
0
votes
1
answer
29
views
Use criterion for statistic driven benchmarking of Substrate pallets
Criterion version 0.4.0 seems to support WASM builds: https://www.tweag.io/blog/2022-03-03-criterion-rs/#first-class-wasm-support
Can we us it to benchmark extrinsics of Substrate pallets?
- 565
0
votes
0
answers
20
views
How do i generate signatures in benchmarking.rs? [duplicate]
I am trying to generate an sr25519 signature within benchmarks. But i am facing an error that doesn't make sense to me.
let sig: T::Signature = acc_pair.sign(&schema.encode()); | -...
- 69
3
votes
2
answers
59
views
Using Autogenerated Weights
I'm wondering about weights. The docs are quite outdated here and I do not understand what is going on. I've managed to create benchmarks and generate a weight.rs.
https://github.com/paritytech/...
- 284
1
vote
1
answer
72
views
What reference hardware can a parachain use to benchmark & generate weights?
The 2022 tracking issue for Substrate benchmarks mentions that Polkadot uses this reference hardware of this specification to generate weights:
Reference Hardware
The transaction weights in Polkadot ...
Nuke♦
- 1,802
0
votes
0
answers
19
views
How can I get time in ms for executing pallet calls?
I'm trying to measure the time a pallet call takes, in milliseconds.
So I can compare for different setups the call 'impact', for example use different numbers of collators, and gather useful data to ...
0
votes
1
answer
60
views
How to decrease time it takes to benchmark?
I have the following benchmark that takes a very long time to complete. What measures can I take to improve the time it takes to run the benchmark without losing its effectiveness?
on_initialize {
...
- 736
2
votes
1
answer
61
views
Benchmark weights output
Can someone please explain the following output:
Why did the weights generate a zero input?
Weight::from_ref_time(0 as u64)
What is the fastest way to eyeball this and conclude whether these weights ...
- 736
0
votes
1
answer
19
views
What does #[skip_meta] do?
I notice some benchmarks in Frame have #[skip_meta].
What does this macro do?
Please explain when one should use this macro when benchmarking.
- 736
0
votes
0
answers
22
views
How to check weather calculated weight are applied to the relevant function? [duplicate]
Normally, a developer write pallet public functions and assigning some hard coded weight to the function. Later on when everything is locked and development of pallet is complete. After that, the ...
- 633
2
votes
1
answer
65
views
How to benchmark functions with multiple logical paths? [duplicate]
The purpose of a benchmarking functions is to find the resources consumed by a functions in the worst case scenario. Is there any tools/methodlogy available so that , we can find the most resource ...
- 633
1
vote
1
answer
64
views
How to test my bench-marking functions?
I have written benchmarking functions from my pallet. How I can test these benchmarking functions please?
- 633
2
votes
1
answer
55
views
why are complexity parameter or accounts seeding unnecessary for benchmarking transfers?
I am trying to understand why there is no complexity parameter on transfer benchmarks.
The substrate benchmarks documentation states:
For benchmarks, you want to execute the most computationally ...
- 736
2
votes
1
answer
64
views
Benchmark extra macro
What does the #[extra] macro do in benchmarks?
I was looking at how benchmarks are being implemented on the Balances pallet. I notice the following benchmarks:
benchmarks_instance_pallet {
transfer {...
- 736
2
votes
1
answer
38
views
Benchmarking pallet code which is not an extrinsic
I am currently writing code for the hooks on_initialize and on_idle. They share functionality in the form of 2 methods, each using these methods multiple times. I would like to accurately benchmark ...
- 183
1
vote
2
answers
87
views
How to get the current weights of a chain for all extrinsics to verify all benchmarks has been integrated?
I have just ran benchmarks and updated all the weights of the pallets in my parachain... or have I?
How can I verify that all my changes "bite"?
Is there a way to check the weights of a ...
- 929
0
votes
0
answers
36
views
Howto customize Babe and Grandpa default weights
It seems polkadot and kusama runtimes both which use babe and grandpa use the default weights for those pallets. The weights are handcrafted ref
In default_weights.rs and default_weights.rs
The ...
0
votes
2
answers
87
views
How to set timestamp in benchmarks?
Is it possible to use std for benchmarks? If not, how can I get timestamps in a benchmark?
I tried using this but it didn't compile:
#[cfg(feature = "runtime-benchmarks")] extern crate std;...
- 510
3
votes
1
answer
43
views
Block number in the benchmark code is 0, but is 1 in the running code
If I'm running a benchmark and I get the current block number from inside the benchmark code (i.e. in the setup before I actually call the call I'm trying to benchmark) using frame_system::Pallet::<...
- 268