feat: add extra sstore benchmark cases #1774
Conversation
LouisTsai-Csie
added
A-test-benchmark
|
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## forks/osaka #1774 +/- ##
============================================
Coverage 86.08% 86.08%
============================================
Files 743 743
Lines 44072 44072
Branches 3891 3891
============================================
Hits 37938 37938
Misses 5656 5656
Partials 478 478
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
spencer-tb
left a comment
spencer-tb
left a comment
There was a problem hiding this comment.
Thanks some comments! :)
|
|
||
| assert len(setup) - 1 == start_marker | ||
| assert len(setup) + len(loop) == end_marker | ||
| print(f"setup: {len(setup)}, loop: {len(loop)}, cleanup: {len(cleanup)}") |
There was a problem hiding this comment.
Should we remove the print?
| print(f"setup: {len(setup)}, loop: {len(loop)}, cleanup: {len(cleanup)}") |
| if contract_idx == 0: | ||
| contract_gas_limit += gas_remainder |
There was a problem hiding this comment.
Why do we give the remainder to the first contract? Does the last contract make more sense?
| for contract_idx in range(num_contracts): | ||
| initial_storage = Storage() | ||
|
|
||
| start_slot = contract_idx * slot_count |
There was a problem hiding this comment.
I think using slot_count instead of slots_per_contract creates storage gaps (for example contract 0 uses slots 0-9, contract 1 uses 100-109, contract 2 uses 200-209). With the latter we get, 0-9, 10-19, 20-29 etc
| start_slot = contract_idx * slot_count | |
| start_slot = contract_idx * slot_per_contract |
| @pytest.mark.parametrize("slot_count", [50, 100]) | ||
| @pytest.mark.parametrize("use_access_list", [True, False]) | ||
| @pytest.mark.parametrize( | ||
| "contract_size", | ||
| [ | ||
| pytest.param(0, id="just_created"), | ||
| pytest.param(1024, id="small"), | ||
| pytest.param(12 * 1024, id="medium"), | ||
| pytest.param(24 * 1024, id="xen"), | ||
| ], | ||
| ) | ||
| @pytest.mark.parametrize("sloads_before_sstore", [True, False]) | ||
| @pytest.mark.parametrize("num_contracts", [1, 5, 10]) | ||
| @pytest.mark.parametrize( | ||
| "initial_value,write_value", | ||
| [ | ||
| pytest.param(0, 0, id="zero_to_zero"), | ||
| pytest.param(0, 0xDEADBEEF, id="zero_to_nonzero"), | ||
| pytest.param(0xDEADBEEF, 0, id="nonzero_to_zero"), | ||
| pytest.param(0xDEADBEEF, 0xBEEFBEEF, id="nonzero_to_nonzero"), | ||
| ], | ||
| ) |
There was a problem hiding this comment.
This will create 384 seperate tests. Do we need this many here? Maybe we should remove some parameterization. I say this with the benchmark release process taking a long time in mind.
Maybe contract size and num contracts can have one element removed from each.
If these are required in your opinion please keep them. Just a thought.
3 participants
🗒️ Description
Add extra SSTORE benchmark cases, more details linked in the issue #1755 .
🔗 Related Issues or PRs
Issue #1755
✅ Checklist
toxchecks to avoid unnecessary CI fails, see also Code Standards and Enabling Pre-commit Checks:uvx tox -e statictype(scope):.mkdocs servelocally and verified the auto-generated docs for new tests in the Test Case Reference are correctly formatted.@ported_frommarker.Cute Animal Picture