tests(benchmark): expand worst case modexp benchmarks

[LouisTsai-Csie]

🗒️ Description

The test cases are divided into two parts: the first tests the upper bound of modexp based on EIP-7823, and the second ports cases from Nethermind. For the latter one, I’ve added a reference to each individual case.

Script for converting the test case: https://gist.github.com/648207a8651d3319ea6dd06e9c977422.git

This PR ports the modexp test cases from Nethermind, you can find the corresponding code here.

I wrote a script to generate test cases, this is my process:

1. I port all the test case from Nethermind's codebase to this text file.
2. Then I use this script to create the correct format we use in EEST, please put your script and the text file in the same folder and run the script: python script.py
3. It will create a file: output.txt and this is the result.

🔗 Related Issues

Issue #1453

✅ Checklist

• All: Set appropriate labels for the changes.
• Tests: Included the type and version of evm t8n tool used to locally execute test cases: e.g., ref with commit hash or geth 1.13.1-stable-3f40e65.
• Tests: Ran mkdocs serve locally and verified the auto-generated docs for new tests in the Test Case Reference are correctly formatted.

[LouisTsai-Csie]

@kevaundray I’ve checked with the Nethermind team regarding the ported test. Could you help confirm whether we can change the ending from "00" to "02" for the even case?

Based on our previous discussion, using "00" as the ending might get stripped by calldata = bytes(mod_exp_input).rstrip(b"\x00"), which could cause issues.

Example:

execution-spec-tests/tests/zkevm/test_worst_compute.py

Line 509 in ead99d1

modulus=7 * "ff" + "00",

execution-spec-tests/tests/zkevm/test_worst_compute.py

Line 815 in ead99d1

calldata = bytes(mod_exp_input).rstrip(b"\x00")

[marioevz]

#1804 got merged! Please rebase the branch and move the tests to the appropriate folder.

[marioevz]

@kevaundray I’ve checked with the Nethermind team regarding the ported test. Could you help confirm whether we can change the ending from "00" to "02" for the even case?

Based on our previous discussion, using "00" as the ending might get stripped by calldata = bytes(mod_exp_input).rstrip(b"\x00"), which could cause issues.

Example:

execution-spec-tests/tests/zkevm/test_worst_compute.py

Line 509 in ead99d1

modulus=7 * "ff" + "00",

execution-spec-tests/tests/zkevm/test_worst_compute.py

Line 815 in ead99d1

calldata = bytes(mod_exp_input).rstrip(b"\x00")

I don't think this is a problem because modexp is supposed to right-pad the input with zeros if the sum of the lengths provided is greater than the calldata input.

It might be better to rollback this change IMO.

[marioevz]

I could not make the conversion script work or rather I don't know what the inputs should be, could you share more info on the input data?

[LouisTsai-Csie]

Thanks @marioevz, I've updated my script and the description, you could find the details in the PR description now.