Create a prepareTxScripts function for profiling scripts #721

ch1bo · 2023-02-10T15:04:44Z

This gets ahold of all the applied plutus scripts in a transaction to profile them externally by returning the fully applied, flat-encoded UPLC programs.

This can be seen as the acquiring an executable script step of the plutus-tx profiling guide.

Thoughts when creating this and some usage tips are in the logbook.

With this profiles and flame-graphs of memory/cpu usage of our hydra scripts like this can be made:

Created a page in /benchmarks on our website, explaining steps to profile scripts in the hydra repository and link it from contribution guidelines.

Created an issue to re-expose mkTermToEvaluate in latest plutus version IntersectMBO/plutus#5133

github-actions · 2023-02-10T15:20:59Z

Transactions Costs

Sizes and execution budgets for Hydra protocol transactions. Note that unlisted parameters are currently using arbitrary values and results are not fully deterministic and comparable to previous runs.

Metadata
Generated at	2023-02-13 14:21:00.842196398 UTC
Max. memory units	14000000
Max. CPU units	10000000000
Max. tx size (kB)	16384

Script summary

Name	Hash	Size (Bytes)
νInitial	7bb44e248f92ae5d2c4c744244b469d59a5bfa8a5d8ce9b6b27e5750	5530
νCommit	70c545fc894ada81ad46715bac1a11fa755b3e3a1d94f03254a4d397	2473
νHead	c1fa50bb6fb1b7f84f4b103bcfb6f71eaeb4de93ddcab99729c122fb	9928

Cost of Init Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	5048	9.94	3.92	0.48
2	5253	13.93	5.52	0.53
3	5458	15.21	5.99	0.56
5	5869	19.50	7.65	0.62
10	6893	28.74	11.20	0.77
46	14276	99.82	38.58	1.86

Cost of Commit Transaction

Currently only one UTxO per commit allowed (this is about to change soon)

UTxO	Tx size	% max Mem	% max CPU	Min fee ₳
1	633	20.95	8.45	0.41

Cost of CollectCom Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	13137	22.07	8.83	0.97
2	13559	37.72	15.25	1.16
3	13840	54.37	22.14	1.36
4	14262	75.56	30.95	1.61
5	14656	99.83	41.08	1.90

Cost of Close Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	10534	16.28	6.99	0.80
2	10733	18.42	8.22	0.83
3	10864	19.41	8.66	0.85
5	11228	22.86	10.63	0.91
10	12052	29.44	14.33	1.02
47	15133	63.02	29.34	1.54

Cost of Contest Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	10571	21.44	8.95	0.85
2	10745	23.24	9.87	0.88
3	10901	24.49	10.57	0.90
5	11265	28.87	12.89	0.97
10	12081	37.49	17.32	1.11
35	16288	81.50	40.35	1.82

Cost of Abort Transaction

Some variation because of random mixture of still initial and already committed outputs.

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	15092	28.55	12.07	1.13
2	15516	48.36	20.76	1.37
3	15343	57.45	24.33	1.46
4	15767	83.32	35.70	1.77

Cost of FanOut Transaction

Involves spending head output and burning head tokens. Uses ada-only UTxO for better comparability.

UTxO	Tx size	% max Mem	% max CPU	Min fee ₳
1	14780	10.10	4.40	0.91
2	14950	12.63	5.68	0.95
3	14987	14.16	6.56	0.97
5	14993	16.87	8.19	1.01
10	15304	25.26	12.89	1.12
20	15599	40.09	21.53	1.32
42	16323	73.60	40.90	1.76

github-actions · 2023-02-10T15:25:52Z

Test Results

285 tests ±0 279 ✔️ ±0 18m 19s ⏱️ + 4m 54s
  97 suites ±0     6 💤 ±0
    5 files ±0     0 ❌ ±0

Results for commit 996d9f0. ± Comparison against base commit dc6b3ef.

♻️ This comment has been updated with latest results.

ghost

Document any hydra-specific steps needed for this? where?

that would be great, I think the documentation would be a good place for that. In general, we should move all the technical stuff that's in the wiki in the docs for more exposure and better tracking, this includes coding conventions, build guidelines, releases strategy, etc.

How to ensure these functions stay exposed at the ledger/plutus-core?

Not sure what you mean here, we don't have much control over the plutus or ledger codebase so if they decide to drop support in some future version, the best we can do is try to be warned in advance and yell.

Add another hydra-specific function to identify scripts by matching hashes from hydra-plutus?

That would be good but probably orthogonal to this PR's feature?

Something I would have added to this PR is calling this function from the inspect-scripts executable for all scripts we have. Perhaps even add a specific CI step to automatically generate a profile from saturated scripts?

ch1bo · 2023-02-10T17:52:43Z

calling this function from the inspect-scripts executable for all scripts we have

This is not possible as the scripts themselves are missing concrete datum, redeemer, and script context arguments. To profile scripts we need all of them. Hence, this function which extracts them from the Tx.

ghost · 2023-02-10T18:06:04Z

calling this function from the inspect-scripts executable for all scripts we have

This is not possible as the scripts themselves are missing concrete datum, redeemer, and script context arguments. To profile scripts we need all of them. Hence, this function which extracts them from the Tx.

I know, but what prevents us from constructing a "Healthy Tx" and extract the fully applied scripts from this tx?

ch1bo · 2023-02-13T07:50:08Z

I know, but what prevents us from constructing a "Healthy Tx" and extract the fully applied scripts from this tx?

That would be possible, but we would need to put something like that into the hydra-node package. Definitely out of scope for this PR I'd say though.

ghost · 2023-02-13T07:51:49Z

Definitely out of scope for this PR I'd say though.
Yes!

This gets ahold of all the applied plutus scripts in a transaction to profile them externally by returning the fully applied, flat-encoded UPLC programs.

v0d1ch

nice!

ghost

Some minor suggestions, feel free to ignore.

ghost · 2023-02-14T08:17:38Z

docs/benchmarks/profiling.md

+
+# Profiling Hydra scripts
+
+This is a quick tutorial how to profile Hydra scripts and is intended for


typos:

a quick tutorial -> a quick tutorial on

and is intended -> , and it is intended

ghost · 2023-02-14T08:18:13Z

docs/benchmarks/profiling.md

+This is a quick tutorial how to profile Hydra scripts and is intended for
+contributors to the `hydra-node`.
+
+On every PR and also for the latest `master`, we do compute typical transaction


suggestion: remove transaction

ghost · 2023-02-14T08:20:50Z

docs/benchmarks/profiling.md

+-- [...]
+```
+
+After running the corresponding code (`tx-cost` in our example), we will be left


suggestion:

running tx-cost executable in our example

ghost · 2023-02-14T08:21:42Z

docs/benchmarks/profiling.md

+
+Unfortunately it's quite hard to tell them apart, but script sizes should help
+in telling the big `vHead` script apart from the smaller `vCommit` script. In
+the profile, names of original `plutus-tx` functions will be retained so that


suggestion: so that -> which

ghost · 2023-02-14T08:22:23Z

docs/benchmarks/profiling.md

+Unfortunately it's quite hard to tell them apart, but script sizes should help
+in telling the big `vHead` script apart from the smaller `vCommit` script. In
+the profile, names of original `plutus-tx` functions will be retained so that
+should make it clear at the latest.


suggestion: at the latest -> which is which?

ghost · 2023-02-14T08:23:15Z

docs/benchmarks/profiling.md

+upstream projects and use your distribution's package manager):
+
+```
+nix shell nixpkgs#flamegraph github:input-output-hk/plutus#x86_64-linux.plutus.library.plutus-project-924.hsPkgs.plutus-core.components.exes.traceToStacks github:input-output-hk/plutus#x86_64-linux.plutus.library.plutus-project-924.hsPkgs.plutus-core.components.exes.uplc


question: why 924? Are we not still using ghc 8.10.7?

Plutus is already supporting 9.2.4 and their nix environment is parameterized by it.

ch1bo requested review from a user, v0d1ch and ffakenz February 10, 2023 15:04

ghost approved these changes Feb 10, 2023

View reviewed changes

ch1bo force-pushed the ch1bo/prepare-tx-scripts branch from 3d147a5 to c157418 Compare February 13, 2023 12:20

ch1bo added 5 commits February 13, 2023 15:06

Move renderScriptExecutionError into where clause

a608708

Create a prepareTxScripts functions

7e2ba64

This gets ahold of all the applied plutus scripts in a transaction to profile them externally by returning the fully applied, flat-encoded UPLC programs.

Write a hydra-specific tutorial for profiling

1984bf5

Add a link to the profiling tutorial to the footer (for now)

dd71435

Move profiling to benchmark section instead

996d9f0

ch1bo force-pushed the ch1bo/prepare-tx-scripts branch from e1ec0af to 996d9f0 Compare February 13, 2023 14:06

ch1bo requested a review from a user February 13, 2023 14:11

v0d1ch approved these changes Feb 14, 2023

View reviewed changes

ch1bo merged commit 1c4c2aa into master Feb 14, 2023

ch1bo deleted the ch1bo/prepare-tx-scripts branch February 14, 2023 08:17

ghost approved these changes Feb 14, 2023

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Create a prepareTxScripts function for profiling scripts #721

Create a prepareTxScripts function for profiling scripts #721

ch1bo commented Feb 10, 2023 •

edited

Loading

github-actions bot commented Feb 10, 2023 •

edited

Loading

github-actions bot commented Feb 10, 2023 •

edited

Loading

ghost left a comment

ch1bo commented Feb 10, 2023

ghost commented Feb 10, 2023

ch1bo commented Feb 13, 2023

ghost commented Feb 13, 2023

v0d1ch left a comment

ghost left a comment

ghost Feb 14, 2023

ghost Feb 14, 2023

ghost Feb 14, 2023

ghost Feb 14, 2023

ghost Feb 14, 2023

ghost Feb 14, 2023

ch1bo Feb 14, 2023


		# Profiling Hydra scripts

		This is a quick tutorial how to profile Hydra scripts and is intended for

Create a prepareTxScripts function for profiling scripts #721

Create a prepareTxScripts function for profiling scripts #721

Conversation

ch1bo commented Feb 10, 2023 • edited Loading

github-actions bot commented Feb 10, 2023 • edited Loading

Transactions Costs

Script summary

Cost of Init Transaction

Cost of Commit Transaction

Cost of CollectCom Transaction

Cost of Close Transaction

Cost of Contest Transaction

Cost of Abort Transaction

Cost of FanOut Transaction

github-actions bot commented Feb 10, 2023 • edited Loading

Test Results

ghost left a comment

Choose a reason for hiding this comment

ch1bo commented Feb 10, 2023

ghost commented Feb 10, 2023

ch1bo commented Feb 13, 2023

ghost commented Feb 13, 2023

v0d1ch left a comment

Choose a reason for hiding this comment

ghost left a comment

Choose a reason for hiding this comment

ghost Feb 14, 2023

Choose a reason for hiding this comment

ghost Feb 14, 2023

Choose a reason for hiding this comment

ghost Feb 14, 2023

Choose a reason for hiding this comment

ghost Feb 14, 2023

Choose a reason for hiding this comment

ghost Feb 14, 2023

Choose a reason for hiding this comment

ghost Feb 14, 2023

Choose a reason for hiding this comment

ch1bo Feb 14, 2023

Choose a reason for hiding this comment

ch1bo commented Feb 10, 2023 •

edited

Loading

github-actions bot commented Feb 10, 2023 •

edited

Loading

github-actions bot commented Feb 10, 2023 •

edited

Loading