perf: decode logs

[sakulstra]

We have an application that indexes a huge amount of logs and when benchmarking where time is spent, i noticed that the decoding alone takes roughly 10% of the time. Therefore, i started looking into how to improve performance here.

Running against this version locally, seems to improve performance by roughly 25% for us.
Still probably there is more things to optimize.

Opening this more for discussion than to necessarily merge.
Our usecase essentially is: decoding a huge amounts of logs with a limited set of abiEvents (perhaps 20).

As i understand the bench result it's around x3 as fast?
In our realworld example (which does more things than just decoding), these are the results:

Before:

✅ Run 1 completed:
   Decoding time: 11.95s
   Total logs: 364674

After:

✅ Run 1 completed:
   Decoding time: 7.00s 
   Total logs: 364674

[changeset-bot]

🦋 Changeset detected

Latest commit: 68aaf3d

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
viem	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[pkg-pr-new]

Open in StackBlitz

npm i https://pkg.pr.new/viem@4228

commit: 68aaf3d

[jxom]

I would be happy to merge this in if it makes things faster.

[sakulstra]

For us, at least in local development it seems to speed things up quite a bit.

Wondering though if there are other things one could do.
Noble has a new major release: https://github.com/paulmillr/noble-curves/releases not sure if that helps much.
There is a pending pr that might improve things: paulmillr/noble-hashes#126