unshuffleList performance

[dapplion]

unshuffleList() can represent a significant part of an epoch transition. This issue recaps some work to check it performance.

1. Comparison against Lighthouse

• Lighthouse swap or not bench
• Lodestar shuffle bench in a branch

list size	Lightouse	Lodestar
512	254.04 us	1.6034 ms (x6)
16384	6.2046 ms	18.272 ms (x3)
4000000	1.5617 s	4.9690 s (x3)

2. Cost of hashing in shuffle

Changing the hash function with an echo function, estimate the total time spent hashing which is around 80%. Our hashing function could be optimized with native bindings, with a ~1.5-2x improve. With that unshuffleList() cpu time could be reduced by 40%.

3. Profile for the record

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 15 days if no further activity occurs. Thank you for your contributions.

[dapplion]

With the improvements in #2353 that's as much performance as we can squeeze. There are many other areas with higher priority to focus now

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 15 days if no further activity occurs. Thank you for your contributions.

[dapplion]

unshuffleList has never shown up in benchmarks recently. Also thanks to the eager state transition ahead of the epoch start, shaving off some milliseconds won't make a difference here